Ever wondered where we could be?
Satyajeet
Salokhe
Product · Marketing · Innovation
AboutÜber mich ProjectsProjekte SkillsKenntnisse ContactKontakt

Product · Project · Marketing · Innovation Produkt · Projekt · Marketing · Innovation

Satyajeet
Salokhe

M.Eng student in Technology & Innovation Management. Building products at the intersection of strategy, design, and engineering. M.Eng-Student in Technologie- und Innovationsmanagement. Produkte an der Schnittstelle von Strategie, Design und Technik.

Explore my work Meine Arbeit entdecken Or read my life in lyrics → Mein Leben in Lyrics →
v.03 — 2025

"And a new day will dawn for those who stand long."

— Led Zeppelin
Satyajeet Salokhe presenting at HS Harz

Presenting research — HS Harz, 2025 Forschungspräsentation — HS Harz, 2025

Built to go
the distance.
Gemacht für
den langen Weg.

I grew up in Kolhapur — a city that smells of sugarcane and engine oil, where my family ran a spray-painting workshop and craft was never separate from work. That early education in making things with your hands never really left me. It just changed form. Aufgewachsen in Kolhapur — einer Stadt, die nach Zuckerrohr und Motoröl riecht, wo meine Familie eine Spritzlackier-Werkstatt betrieb und Handwerk nie von Arbeit zu trennen war. Diese frühe Erziehung im Erschaffen mit den Händen hat mich nie wirklich verlassen. Sie hat nur die Form gewechselt.

Pune followed — engineering, late nights, a degree from D.Y. Patil, and two years learning how real businesses actually run. I coordinated triathlon events, managed supply chains, understood that strategy without execution is just noise. Then Germany called, and I answered. Dann Pune — Ingenieurwesen, späte Nächte, ein Abschluss am D.Y. Patil, zwei Jahre damit, zu verstehen, wie echte Unternehmen wirklich funktionieren. Triathlon-Events koordiniert, Lieferketten gemanagt, verstanden, dass Strategie ohne Umsetzung nur Lärm ist. Dann rief Deutschland, und ich antwortete.

Now at Hochschule Harz, building at the intersection of product, marketing and innovation. Fuelled by football, history, and the kind of music that reminds you that some things are worth the long road — Zeppelin, GN'R, the whole canon. Jetzt an der Hochschule Harz, an der Schnittstelle von Produkt, Marketing und Innovation. Angetrieben von Fußball, Geschichte und Musik, die einem daran erinnert, dass manche Dinge den langen Weg wert sind — Zeppelin, GN'R, der ganze Kanon.

LocationStandort Wernigerode, DE
EducationAusbildung M.Eng — Hochschule Harz
Expected 2.2 · OngoingAngestrebt 2,2 · Laufend
BackgroundHintergrund Kolhapur, IndiaKolhapur, Indien
LanguagesSprachen English C2 · Deutsch B1Englisch C2 · Deutsch B1
CertificatesZertifikate Lean Six Sigma Yellow Belt
Excel · Project Management
Lean Six Sigma Yellow Belt
Excel · Projektmanagement
FocusSchwerpunkt Product · Project
Marketing · Innovation
Produkt · Projekt
Marketing · Innovation

Things I've built
& explored
Was ich entwickelt
& erforscht habe

001

Sustainable Product Strategy — Recycled RubberNachhaltige Produktstrategie — Recycelter Gummi

End-to-end product strategy: user persona research, stakeholder analysis, requirements definition, and concept pitching for an eco-friendly rubber product line.End-to-End-Produktstrategie: Nutzer-Persona-Forschung, Stakeholderanalyse, Anforderungsdefinition und Konzeptpitch für eine umweltfreundliche Gummiproduktlinie.

Market ResearchMarktforschung User PersonasNutzer-Personas SustainabilityNachhaltigkeit

002

Smart Necklace Product ConceptProduktkonzept: Intelligente Halskette

Led an Agile sprint for concept validation and feedback-driven design iteration. Produced a digital marketing video in DaVinci Resolve and Canva.Leitete einen Agile-Sprint zur Konzeptvalidierung. Produzierte ein digitales Marketingvideo in DaVinci Resolve und Canva.

Agile Sprint Marketing VideoMarketingvideo Concept DesignKonzeptdesign

003

YOLO-Based Drone RecognitionYOLO-basierte Drohnenerkennung

Real-time object detection system achieving >90% precision — reducing false positives by 25% and improving detection speed by 35% in an Industry 4.0 environment.Echtzeit-Objekterkennung mit >90% Präzision — 25% weniger Falschpositive, 35% schnellere Erkennung in einer Industrie-4.0-Umgebung.

YOLOv11 Computer Vision Industry 4.0

004

Idea Engineering — Elbfabrik Collaboration FormatsIdea Engineering — Kollaborationsformate Elbfabrik

Systematic innovation management study: applied 5 Whys root cause analysis, brainstorming, and multi-criteria Benefit Analysis to design 5 collaboration formats for Fraunhofer IFF's Elbfabrik hub.Systematische Innovationsmanagementstudie: 5-Warum-Analyse, Brainstorming und Nutzwertanalyse zur Entwicklung von 5 Kollaborationsformaten für das Fraunhofer IFF Elbfabrik.

Innovation StrategyInnovationsstrategie Benefit AnalysisNutzwertanalyse 5 Whys5-Warum-Methode

005

Paper Review — Reliable User Identification in Social Media During CrisisPaper-Review — Zuverlässige Nutzeridentifikation in sozialen Medien während Krisen

Critical assessment and lecture presentation of a published deep learning paper — evaluated a Bi-LSTM system achieving 94.54% accuracy in classifying disaster tweets, analysing methodology, results, and real-world applicability.Präsentation eines akademischen Papers: Bi-LSTM-System mit 94,54 % Genauigkeit bei der Klassifizierung katastrophenbezogener Tweets mittels VADER-Sentiment-Analyse und K-Means-Clustering.

Bi-LSTM Sentiment AnalysisSentiment-Analyse NLP Deep Learning

006

Technology Assessment — Smart Homes & BuildingsTechnologiebewertung — Smart Homes & Gebäude

Group TAS project comparing smart home innovation strategies for the USA and Thailand. Conducted PESTEL analysis, proposed four innovations, and produced a Technology Radar mapping adoption readiness across AI, IoT, and blockchain tools.Gruppen-TAS-Projekt: PESTEL-Analyse für USA und Thailand, vier Innovationsvorschläge und ein Technology Radar zur Bewertung von KI-, IoT- und Blockchain-Tools.

PESTEL Technology RadarTechnologie-Radar IoT SDGs

How I
work
Wie ich
arbeite

Three modes. Each one is a different kind of problem — and a different set of tools working together to solve it. Drei Arbeitsmodi. Jeder löst eine andere Art von Problem — mit unterschiedlichen Tools, die zusammenwirken.

01

Brief → PrototypeBrief → Prototyp

A stakeholder describes a problem. Within a session I can go from blank page to something tangible enough to react to — using Claude to pressure-test the brief, Figma to make it visual, Canva when it needs to move or present. The gap between "idea" and "thing you can show someone" has collapsed. Ein Stakeholder beschreibt ein Problem. Innerhalb einer Session kann ich von der leeren Seite zu etwas Greifbarem gelangen — Claude zum Hinterfragen des Briefs, Figma zur Visualisierung, Canva wenn es präsentiert werden muss.

Claude ChatGPT Copilot Figma FigJam Canva Notion AI Miro
Design ThinkingDesign Thinking Rapid PrototypingRapid Prototyping User Persona ResearchNutzer-Persona-Forschung Stakeholder PitchingStakeholder-Pitching

02

Complexity → ClarityKomplexität → Klarheit

When a problem is genuinely messy — multiple stakeholders, conflicting inputs, unclear root cause — I reach for structured methods on collaborative canvases. Miro and Conceptboard for mapping and ideation, 5 Whys and Benefit Analysis for working through the fog to a defensible recommendation. This is where I actually enjoy the process most. Bei wirklich unübersichtlichen Problemen — mehrere Stakeholder, widersprüchliche Inputs — nutze ich strukturierte Methoden auf kollaborativen Canvases. Miro und Conceptboard zum Mapping, 5-Warum und Nutzwertanalyse für klare Empfehlungen.

Miro Conceptboard FigJam Jira MS Teams Notion Excel
5 Whys Benefit AnalysisNutzwertanalyse Scrum & SAFe Lean Six Sigma Risk AnalysisRisikoanalyse Innovation Platform DesignInnovationsplattform-Design Idea EngineeringIdeen-Engineering

03

Concept → AudienceKonzept → Publikum

Once the thinking is done, the work isn't. Getting a validated idea in front of the right people requires a different skill — translating it into whatever format they need. A marketing video edited in DaVinci. A data story in Power BI. A product deck in Canva. The tool always serves the audience, not the other way around. Sobald das Denken abgeschlossen ist, ist die Arbeit nicht vorbei. Eine validierte Idee vor das richtige Publikum zu bringen erfordert eine andere Fertigkeit — sie in das jeweils passende Format zu übersetzen.

DaVinci Resolve Power BI Excel Canva Figma Jira & MS Teams Notion
Digital MarketingDigitales Marketing Market ResearchMarktforschung Change ManagementChange Management Tech AdoptionTechnologieadoption PESTEL Technology RadarTechnologie-Radar

Let's connectLass uns vernetzen

Get in touchKontakt aufnehmen

satyajeetsalokhe57
@gmail.com

Open to internships, collaborations
& new opportunities.
Offen für Praktika, Kooperationen
& neue Möglichkeiten.

Find me onlineOnline finden

LinkedIn

Wernigerode, Deutschland
+49 017634415608

Product Strategy · HS HarzProduktstrategie · HS Harz

Sustainable Product Strategy
Recycled Rubber
Nachhaltige Produktstrategie
Recycelter Gummi

A full product strategy project exploring the feasibility and market potential of eco-friendly rubber products — from user persona definition to stakeholder pitching and concept prototyping.Ein vollständiges Produktstrategieprojekt zur Bewertung der Machbarkeit und des Marktpotenzials umweltfreundlicher Gummiprodukte — von der Nutzer-Persona-Definition bis hin zu Stakeholder-Pitching und Konzeptprototyping.

User Personas DefinedDefinierte Nutzer-Personas

Persona 1 — The Eco-Conscious ConsumerPersona 1 — Der umweltbewusste Verbraucher

Urban professional, 25–40, actively chooses sustainable brands. Willing to pay a premium for certified eco-friendly products. Values transparency in materials and production.Urbaner Berufstätiger, 25–40, wählt aktiv nachhaltige Marken. Bereit, einen Aufpreis für zertifizierte Öko-Produkte zu zahlen. Legt Wert auf Transparenz bei Materialien und Produktion.

Persona 2 — The Industrial BuyerPersona 2 — Der industrielle Käufer

Procurement manager in manufacturing, 35–55. Prioritises cost-efficiency and compliance. Interested in recycled materials for regulatory and ESG reporting advantages.Einkaufsleiter in der Industrie, 35–55. Priorisiert Kosteneffizienz und Compliance. Interessiert an recycelten Materialien für regulatorische und ESG-Reporting-Vorteile.

Persona 3 — The Retail PartnerPersona 3 — Der Einzelhandelspartner

Buyer at a mid-sized retail chain, 30–50. Seeks differentiated products with sustainability credentials to meet growing consumer demand and brand positioning goals.Einkäufer bei einer mittelgroßen Einzelhandelskette, 30–50. Sucht differenzierte Produkte mit Nachhaltigkeitsnachweisen, um der wachsenden Verbrauchernachfrage gerecht zu werden.

MethodologyMethodik

  • Competitor analysis — mapped 12 rubber product alternatives across price, eco-certification, and market positioningWettbewerbsanalyse — 12 Gummiprodukt-Alternativen nach Preis, Öko-Zertifizierung und Marktpositionierung kartiert
  • Stakeholder interviews — gathered requirements from potential buyers, suppliers, and end usersStakeholder-Interviews — Anforderungen von potenziellen Käufern, Lieferanten und Endnutzern gesammelt
  • Product requirements document — defined functional, quality, and sustainability specificationsProduktanforderungsdokument — funktionale, qualitative und Nachhaltigkeitsspezifikationen definiert
  • Feasibility assessment — evaluated technical and financial viability of three concept directionsMachbarkeitsbewertung — technische und finanzielle Tragfähigkeit von drei Konzeptrichtungen bewertet
  • Concept pitch — presented validated product concept to stakeholders with business caseKonzeptpitch — validiertes Produktkonzept mit Business Case an Stakeholder präsentiert

Key InsightsWichtige Erkenntnisse

Market GapMarktlücke

Identified an underserved segment of mid-market B2B buyers seeking eco-certified rubber components with competitive pricing — no strong incumbent players at this intersection.Eine unterversorgte Zielgruppe von mittelständischen B2B-Käufern identifiziert, die öko-zertifizierte Gummikomponenten zu wettbewerbsfähigen Preisen suchen.

Value PropositionWertversprechen

Recycled rubber products positioned as both a cost-reduction lever and an ESG compliance tool — giving buyers dual justification for procurement decisions.Recycelte Gummiprodukte als Kostensenkungshebel und ESG-Compliance-Tool positioniert — doppelte Rechtfertigung für Beschaffungsentscheidungen.

Outcome & DeliverablesErgebnis & Liefergegenstände

  • 3 fully defined user personas with behavioural profiles and buying triggers3 vollständig definierte Nutzer-Personas mit Verhaltensprofile und Kaufauslöser
  • Competitive landscape map across 12 productsWettbewerbslandschaftskarte für 12 Produkte
  • Product Requirements Document (PRD) — 3 concept directionsProduktanforderungsdokument (PRD) — 3 Konzeptrichtungen
  • Stakeholder pitch deck with feasibility assessment and business caseStakeholder-Pitch-Deck mit Machbarkeitsbewertung und Business Case

Modules AppliedAngewandte Module

Product DevelopmentProduktentwicklung Product MarketingProduktmarketing Innovation StrategyInnovationsstrategie

Agile Sprint · Marketing Production · HS HarzAgile-Sprint · Marketingproduktion · HS Harz

Smart Necklace
Product Concept
Intelligente Halskette
Produktkonzept

A concept product developed through a structured Agile sprint — from ideation to validated prototype — culminating in a professional digital marketing video produced in DaVinci Resolve and Canva.Ein Konzeptprodukt, das durch einen strukturierten Agile-Sprint entwickelt wurde — von der Ideenfindung bis zum validierten Prototyp — mit einem professionellen Marketingvideo in DaVinci Resolve und Canva.

Project OverviewProjektübersicht

GoalZiel

Design and validate a smart wearable necklace concept through rapid Agile iteration, gathering user feedback at each stage to refine the product vision.Entwurf und Validierung eines intelligenten Halsketten-Konzepts durch schnelle Agile-Iteration mit Nutzerfeedback in jeder Phase.

OutcomeErgebnis

Fully iterated product concept backed by user testing — documented in a professional digital marketing video for stakeholder presentation.Vollständig iteriertes Produktkonzept, gestützt durch Nutzertests — dokumentiert in einem professionellen Marketingvideo für die Stakeholder-Präsentation.

Tools & MethodsTools & Methoden

  • Agile Sprint — structured iteration with defined sprints and retrospectivesAgile-Sprint — strukturierte Iteration mit definierten Sprints und Retrospektiven
  • User feedback sessions — validation and design refinement at each stageNutzerfeedback-Sessions — Validierung und Designverfeinerung in jeder Phase
  • DaVinci Resolve — video editing and colour gradingDaVinci Resolve — Videobearbeitung und Farbkorrektur
  • Canva — graphic design, motion assets, and brand visualsCanva — Grafikdesign, Motion-Assets und Markenvisuals

Sprint ProcessSprint-Prozess

01 — Ideation01 — Ideenfindung

Defined target user, core features, and value proposition through team brainstorming and initial market scan. Set sprint goals and acceptance criteria.Zielnutzer, Kernfunktionen und Wertversprechen durch Team-Brainstorming und Marktanalyse definiert. Sprint-Ziele und Akzeptanzkriterien festgelegt.

02 — Prototype & Test02 — Prototyp & Test

Built low-fidelity concept mockups and ran user feedback sessions. Iterated the design across multiple rounds, incorporating insights each cycle.Low-Fidelity-Mockups erstellt und Nutzerfeedback-Sessions durchgeführt. Design über mehrere Runden iteriert und Erkenntnisse jedes Zyklus einbezogen.

03 — Produce & Present03 — Produktion & Präsentation

Produced a professional digital marketing video presenting the validated concept. Edited in DaVinci Resolve; graphics and motion in Canva.Professionelles Marketingvideo produziert. Bearbeitet in DaVinci Resolve; Grafiken und Motion-Assets in Canva.

Marketing VideoMarketingvideo

Product marketing video hosted on Google DriveProduktmarketingvideo auf Google Drive gehostet

Watch Video →Video ansehen →
Smart Necklace — Digital Marketing Video · DaVinci Resolve & CanvaIntelligente Halskette — Digitales Marketingvideo · DaVinci Resolve & Canva

Research Project · HS HarzForschungsprojekt · HS Harz

YOLO-Based Drone &
Robot Recognition in
Ongoing Video
YOLO-basierte Drohnen- &
Robotererkennung in
laufenden Videos

A YOLOv11-based AI model trained to detect Robomaster TT drones and iRobot ground vehicles — exploring whether image-based recognition alone is sufficient for reliable real-time identification in Industry 4.0 environments.Ein YOLOv11-basiertes KI-Modell zur Erkennung von Robomaster TT-Drohnen und iRobot-Bodenfahrzeugen — untersucht, ob bildbasierte Erkennung allein für zuverlässige Echtzeiterkennung in Industrie-4.0-Umgebungen ausreicht.

ModelModellYOLOv11 · Roboflow
mAP72.6%
PrecisionPräzision100%
Recall66.4%
ConfidenceKonfidenz93%
Co-authorCo-AutorVijay Potekar

Research QuestionsForschungsfragen

Q1

Can robots and drones be consistently identified in onboard video images?Können Roboter und Drohnen in Onboard-Videobildern zuverlässig identifiziert werden?

Q2

Are images alone sufficient, or is additional sensor data required for reliable detection?Reichen Bilder allein aus oder werden zusätzliche Sensordaten für eine zuverlässige Erkennung benötigt?

Tools & StackTools & Technologien

  • YOLOv11 via Roboflow — primary detection modelYOLOv11 via Roboflow — primäres Erkennungsmodell
  • Robomaster TT Drone + iRobot — hardware subjectsRobomaster TT Drohne + iRobot — Hardware-Objekte
  • High-performance GPUs — model trainingHochleistungs-GPUs — Modelltraining
  • 244 images — 231 train / 8 valid / 5 test244 Bilder — 231 Training / 8 Validierung / 5 Test
  • 1200×1200px · Batch size 249

MethodologyMethodik

01 — Data01 — Daten

Indoor and outdoor images at 2m–50m distance, multiple angles and lighting conditions.Innen- und Außenbilder in 2m–50m Abstand, verschiedene Winkel und Licht.

02 — Processing02 — Verarbeitung

Resizing, noise reduction, and augmentation — flips, rotations, brightness, grayscale, color jitter.Größenanpassung, Rauschreduzierung, Augmentierung — Spiegelungen, Rotationen, Helligkeit.

03 — Annotation03 — Annotation

Manual bounding box annotation via Roboflow with multi-round quality checks.Manuelle Bounding-Box-Annotation via Roboflow mit Qualitätsprüfungen.

04 — Training04 — Training

YOLOv11 retrained on drone-specific dataset. Batch 249 selected for optimal convergence.YOLOv11 auf drohnenspezifischem Datensatz nachtrainiert. Batch 249 für optimale Konvergenz.

Results & FindingsErgebnisse

100% precision — every drone flagged was correctly identified, zero false positives. Critical for autonomous navigation where false detections cause unnecessary evasive actions.100% Präzision — jede erkannte Drohne korrekt identifiziert, null Falschpositive. Entscheidend für autonome Navigation.

Recall at 66.4% highlights room to improve detection in complex backgrounds and occlusion scenarios.Recall von 66,4% zeigt Verbesserungspotenzial bei komplexen Hintergründen und Verdeckung.

Performance MetricsLeistungskennzahlen

PrecisionPräzision100%
mAP72.6%
Recall66.4%
ConfidenceKonfidenz93%

Photo GalleryFotogalerie

Robomaster TT Drone lab setup
Robomaster TT Drone outdoor flight test
Robomaster TT — Lab setupRobomaster TT — Laboraufbau
Robomaster TT — Outdoor flight testRobomaster TT — Außenflugtest

Future DirectionsZukünftige Richtungen

Dataset ExpansionDatensatzerweiterung

Broader scenarios and more drone types to improve generalization.Breitere Szenarien und mehr Drohnentypen zur Verbesserung der Generalisierung.

Sensor FusionSensorfusion

Combining camera data with LiDAR or radar for improved recall in occluded conditions.Kombination von Kameradaten mit LiDAR oder Radar für besseren Recall bei Verdeckung.

Real-world DeploymentRealer Einsatz

Testing in live operational environments with actual drone systems.Tests in realen Umgebungen mit echten Drohnensystemen.

Innovation Management · HS Harz · Group 5Innovationsmanagement · HS Harz · Gruppe 5

Idea Engineering —
Elbfabrik Collaboration Formats
Idea Engineering —
Elbfabrik Kollaborationsformate

A systematic innovation management case study applying structured problem analysis (5 Whys), divergent idea generation (Brainstorming), and objective evaluation (Benefit Analysis) to design five collaboration formats for Fraunhofer IFF's Elbfabrik innovation hub in Magdeburg.Eine systematische Innovationsmanagementstudie mit strukturierter Problemanalyse (5-Warum), divergenter Ideengenerierung (Brainstorming) und objektiver Bewertung (Nutzwertanalyse) zur Entwicklung von fünf Kollaborationsformaten für das Fraunhofer IFF Elbfabrik.

Problem Analysis — 5 WhysProblemanalyse — 5-Warum-Methode

Core Problem IdentifiedKernproblem identifiziert

Lack of effective cross-disciplinary collaboration within Elbfabrik's innovation ecosystem — traced through five layers to its root cause: no structured formats enabling knowledge exchange across academia, industry, and startups.Mangelnde fachübergreifende Zusammenarbeit im Innovationsökosystem des Elbfabrik — auf die Wurzelursache zurückgeführt: keine strukturierten Formate für den Wissensaustausch.

Root Cause ChainUrsachenkette

Seamless communication gaps → limited shared goals → barriers to knowledge exchange → skill mismatches → no structured collaboration formatsKommunikationslücken → fehlende gemeinsame Ziele → Barrieren beim Wissensaustausch → Kompetenzlücken → keine strukturierten Formate

Methodology StackMethodenstack

  • 5 Whys — structured root cause identification beyond surface symptoms5-Warum-Methode — strukturierte Ursachenanalyse jenseits von Oberflächensymptomen
  • Brainstorming — open, fast-paced idea generation for high volume of diverse formatsBrainstorming — offene, schnelle Ideengenerierung für vielfältige Formate
  • Benefit Analysis — multi-criteria scoring across Feasibility, Scalability, Impact, and CostNutzwertanalyse — Mehrkriterien-Bewertung nach Machbarkeit, Skalierbarkeit, Wirkung und Kosten
  • Group 5 — interdisciplinary collaboration between three M.Eng studentsGruppe 5 — interdisziplinäre Zusammenarbeit zwischen drei M.Eng-Studierenden

Benefit Analysis Results — Ranked FormatsErgebnisse der Nutzwertanalyse — Bewertete Formate

01 · 22/25

Innovation Boot Camp

Intensive 1–2 week hackathon-style programme. Highest ranked.

02 · 20/25

Open House Exhibition

Co-creation fair modelled on Hannover Messe.

03 · 20/25

Startup Partnership

Structured startup co-development pathway.

04 · 19/25

Exchange Program

Cross-disciplinary collaboration initiative.

05 · 18/25

Virtual Labs

Online remote collaboration platform.

Key TakeawayKernaussage

Innovation Boot Camp — #1 RecommendedInnovation Boot Camp — Empfehlung #1

The Boot Camp scored highest (22/25) due to its concentrated structure, cross-disciplinary intensity, and ability to directly address all identified root causes simultaneously. Partners: universities (credit modules), startups (infrastructure access), Elbfabrik (talent pipeline + reputation).Das Boot Camp erzielte die höchste Punktzahl (22/25) durch seine konzentrierte Struktur und die Fähigkeit, alle identifizierten Ursachen gleichzeitig zu adressieren.

DeliverablesLiefergegenstände

  • Root cause analysis via 5 Whys — documented problem chainUrsachenanalyse via 5-Warum-Methode — dokumentierte Problemkette
  • 5 collaboration format proposals with execution strategies5 Kollaborationsformate mit Ausführungsstrategien
  • Multi-criteria Benefit Analysis scoring table (4 criteria, 5 formats)Nutzwertanalyse-Tabelle (4 Kriterien, 5 Formate)
  • Group presentation deck — Hochschule Harz, November 2024Gruppenprä­sentations­deck — Hochschule Harz, November 2024

Modules AppliedAngewandte Module

Idea EngineeringIdea Engineering Innovation ManagementInnovationsmanagement Benefit AnalysisNutzwertanalyse

Deep Learning · NLP · HS Harz · Individual PresentationDeep Learning · NLP · HS Harz · Einzelpräsentation

Deep Learning — Reliable
User Identification During Crisis
Deep Learning — Zuverlässige
Nutzeridentifikation während Krisen

A scientific paper reading and critical assessment exercise — part of a university lecture series on AI research. I reviewed, evaluated, and presented a published paper proposing a Bi-LSTM system for identifying trustworthy social media users during disasters, covering its methodology, dataset, model architecture, and findings.Präsentation eines akademischen Papers zur Identifikation zuverlässiger Social-Media-Nutzer während Katastrophen. Das Bi-LSTM-Modell wurde an 20.000 echten Tweets der Assam-Überschwemmungen 2022 evaluiert und kombiniert NLP, VADER-Sentiment-Analyse und K-Means-Clustering.

Research QuestionForschungsfrage

Core ProblemKernproblem

During crises, social media is flooded with misinformation, fake news, and unreliable sources. How can a system accurately predict user trustworthiness in real time to support emergency response teams?Während Krisen wird Social Media mit Fehlinformationen überflutet. Wie kann ein System die Vertrauenswürdigkeit von Nutzern in Echtzeit vorhersagen?

DatasetDatensatz

20,000 disaster tweets collected via Twitter API using 2022 Assam Flood hashtags — reduced to 16,600 after preprocessing. 13,178 user profiles analysed across 380,784 historical tweets.20.000 Katastrophen-Tweets via Twitter API gesammelt (2022 Assam-Überschwemmungen), auf 16.600 nach Vorverarbeitung reduziert. 13.178 Nutzerprofile über 380.784 historische Tweets analysiert.

System ArchitectureSystemarchitektur

  • Tweet preprocessing — tokenization, lemmatization, stop-word removalTweet-Vorverarbeitung — Tokenisierung, Lemmatisierung, Stoppwort-Entfernung
  • User Profile Analysis — engagement metrics, verification status, follower countsNutzerprofilanalyse — Engagement-Metriken, Verifizierungsstatus, Followerzahl
  • User Behavior Analysis — VADER sentiment scoring across past 50 tweets per userNutzerverhaltenanalyse — VADER-Sentiment-Bewertung der letzten 50 Tweets pro Nutzer
  • Bi-LSTM Tweet Classification — relevant vs. irrelevant disaster tweetsBi-LSTM Tweet-Klassifizierung — relevante vs. irrelevante Katastrophen-Tweets
  • K-Means Clustering — combines all scores to surface highly reliable usersK-Means-Clustering — kombiniert alle Scores zur Identifikation zuverlässiger Nutzer

ML Model Performance — Accuracy over 10 EpochsML-Modell-Performance — Genauigkeit über 10 Epochen

CNN

Peak: 85.3%

Captures spatial text patterns but lacks sequential context — lowest performer of the three models.

LSTM

Peak: 92.5%

Handles long-term sequential dependencies well, strong improvement over CNN across all epochs.

Bi-LSTM · Best

Peak: 94.54%

Processes input in both directions, capturing full sentence context — highest accuracy, selected model.

Key FindingsKernerkenntnisse

Sentiment Breakdown (8,235 users after filtering)Sentiment-Verteilung (8.235 Nutzer nach Filterung)

Positive: 2,178 users · Negative: 11,000 users · Neutral: 3,422 eliminated. Negative sentiment dominated — consistent with crisis communication patterns where urgency and distress are prevalent.Positiv: 2.178 · Negativ: 11.000 · Neutral: 3.422 eliminiert. Negatives Sentiment dominierte — konsistent mit Krisenkommunkationsmustern.

Deliverables & ContextLiefergegenstände & Kontext

  • Academic paper review and structured presentation — 24 slidesAkademische Paper-Auswertung und strukturierte Präsentation — 24 Folien
  • System architecture walkthrough — pipeline from raw tweets to reliable user outputSystemarchitektur-Walkthrough — Pipeline von rohen Tweets zum zuverlässigen Nutzeroutput
  • ML model comparison — CNN vs. LSTM vs. Bi-LSTM across 10 epochsML-Modellvergleich — CNN vs. LSTM vs. Bi-LSTM über 10 Epochen
  • Assessment score: 25/30 — strong marks for novelty, quality, and structureBewertung: 25/30 — starke Noten für Neuheit, Qualität und Struktur

Modules AppliedAngewandte Module

Deep Learning NLP Bi-LSTM Sentiment AnalysisSentiment-Analyse

Technology Assessment · HS Harz · Group 5Technologiebewertung · HS Harz · Gruppe 5

Technology Assessment —
Smart Homes & Buildings
Technologiebewertung —
Smart Homes & Gebäude

A comparative Technology Assessment study examining smart home innovation opportunities in the USA and Thailand. The project combined PESTEL analysis, four targeted innovation proposals, SDG alignment, and a Technology Radar evaluating AI, IoT, and blockchain tools across adoption readiness levels.Eine vergleichende Technologiebewertungsstudie zu Smart-Home-Innovationen in den USA und Thailand. PESTEL-Analyse, vier Innovationsvorschläge, SDG-Zuordnung und Technology Radar zur Bewertung von KI-, IoT- und Blockchain-Tools.

PESTEL HighlightsPESTEL-Highlights

USA

Government-backed smart infrastructure investment, evolving cybersecurity legislation, and post-COVID behavioural shifts driving demand for home automation and energy-efficient systems.Staatlich geförderter Infrastrukturausbau, sich entwickelnde Cybersicherheitsgesetze und post-COVID-Verhaltensänderungen treiben die Nachfrage nach Heimautomatisierung an.

Thailand

Rapid urbanisation fuelling smart home market growth, strong environmental awareness shaping consumer preferences, and GDP expansion creating investment capacity for IoT infrastructure.Schnelle Urbanisierung treibt das Smart-Home-Marktwachstum voran, starkes Umweltbewusstsein prägt Verbraucherpräferenzen, BIP-Wachstum schafft IoT-Investitionskapazität.

Innovations ProposedVorgeschlagene Innovationen

  • Smart Thermostat (USA) — voice/gesture control, geofencing, predictive HVAC maintenanceSmart Thermostat (USA) — Sprach-/Gestensteuerung, Geofencing, vorausschauende HVAC-Wartung
  • Solar Panels with AI Energy Management (USA) — bifacial panels, dynamic tilt, +30% energy captureSolaranlagen mit KI-Energiemanagement (USA) — bifaziale Panels, dynamische Neigung, +30 % Energieaufnahme
  • Smart Water Supply Management (Thailand) — IoT leak detection, real-time consumption monitoringSmart Wasserversorgung (Thailand) — IoT-Leckerkennung, Echtzeit-Verbrauchsüberwachung
  • Smart Security System (Thailand) — AI predictive alerts, biometric auth, blockchain footage storageSmart-Sicherheitssystem (Thailand) — KI-Prädiktionsalarme, biometrische Authentifizierung, Blockchain-Speicherung

Technology Radar — Smart Security SystemTechnology Radar — Smart-Sicherheitssystem

Adopt

AI/ML Frameworks · Facial Recognition · Voice Recognition · Cloud Computing · IoT Platforms · Python · OpenCV · TensorFlow/PyTorchKI/ML-Frameworks · Gesichtserkennung · Spracherkennung · Cloud Computing · IoT-Plattformen

Trial

Emotion Recognition · Security Analytics Tools · Edge Computing DevicesEmotionserkennung · Sicherheitsanalysetools · Edge-Computing-Geräte

Assess

Blockchain Storage · Blockchain Platform · Blockchain FrameworksBlockchain-Speicherung · Blockchain-Plattform · Blockchain-Frameworks

SDG Links

SDG 7 — Affordable Clean Energy · SDG 11 — Sustainable Cities · SDG 13 — Climate Action · SDG 6 — Clean WaterSDG 7 · SDG 11 · SDG 13 · SDG 6

Key TakeawayKernaussage

Cross-Market StrategyCross-Market-Strategie

Smart home innovation must be context-sensitive. The USA benefits from energy automation and solar integration, while Thailand's priorities around water scarcity and urban security call for distinct, regionally tailored solutions. Collaboration between government, industry, and academia is essential to meet SDG targets.Smart-Home-Innovation muss kontextsensitiv sein. Die USA profitieren von Energieautomatisierung, während Thailands Prioritäten bei Wasserknappheit und urbaner Sicherheit andere Lösungen erfordern.

DeliverablesLiefergegenstände

  • PESTEL analysis for USA and Thailand smart home marketsPESTEL-Analyse für Smart-Home-Märkte in den USA und Thailand
  • 4 targeted innovation proposals with technical specifications4 Innovationsvorschläge mit technischen Spezifikationen
  • Technology Radar for Smart Security System — 4-quadrant adoption mappingTechnology Radar für Smart-Sicherheitssystem — 4-Quadranten-Adoptionskartierung
  • SDG alignment analysis across all proposed innovationsSDG-Ausrichtungsanalyse für alle vorgeschlagenen Innovationen
  • Group 5 — collaborative TAS module project, Hochschule HarzGruppe 5 — kollaboratives TAS-Modulprojekt, Hochschule Harz

Modules AppliedAngewandte Module

PESTEL Technology RadarTechnologie-Radar IoT SDGs
Back
My Life in Lyrics

Ten chapters, three cities, and the songs that were playing when it all happened.

Chapter 01 · Kolhapur
"Everything comes back to you"
This Town · Niall Horan

A city of sugarcane and engine oil, where Saturdays smell like fresh paint from my father's workshop. I grew up watching people make things — really make things — with patience and pride. That never left me. It just changed form.

I was the elder child. The one who organised the game, gathered the crowd, knew everyone on the street by name. Football was my first language — played everywhere, with everyone, until a knee injury quietly closed that door. I didn't grieve it long. I just found other games to master.

Kolhapur gave me something no degree teaches — the instinct to show up, make something real, and take pride in the work. Everything I've called a skill since was already being practised here. I just didn't have the vocabulary for it yet.

Chapter 02 · The Dreamer
"Oh, let the sun beat down upon my face, with stars to fill my dream"
Kashmir · Led Zeppelin

Before any of it, there was just a boy who collected lyrics the way others collected trophies.

I noticed things. The way a crowd moved. The way the right words at the right moment could shift an entire room. The way some people walked into spaces and immediately understood who held the energy and what needed to be said.

I didn't know yet that noticing was a skill. I just knew the sun felt different when you were moving toward something.

Chapter 03 · The Leap
"As I dream about movies they won't make of me when I'm dead"
Bed of Roses · Bon Jovi

Everyone had a plan that looked like everyone else's plan. Stay. Build. Continue. Safe, known, sensible.

I chose Pune. Not because Kolhapur wasn't enough — because I was. Arrived with contacts, curiosity, and a restlessness that couldn't sit still. Built a team. Started volunteering at sports events — first as a pair of hands, then as the person running the room. Organised triathlons, managed logistics, led before anyone asked me to.

That's when I understood something no classroom teaches: execution is the strategy.

Chapter 04 · Cycle Republik
"You're face to face with the man who sold the world"
The Man Who Sold The World · David Bowie

At Cycle Republik I met someone who could sell anything — not through pressure, through conviction. He understood people before he understood products. He saw something in me before I saw it in myself. Handed me a brief, pointed at a crowd, and said — go.

I went. Leveraged every contact I had. Turned events into experiences. Did marketing the way it should always be done — on the ground, in the room, in the moment.

That mentor didn't give me a playbook. He gave me permission to trust my instincts. That was worth more than any framework.

Chapter 05 · The Year
"I put my heart into this game like I opened my chest"
Good Life · G-Eazy & Kehlani

Then came the year I failed. Not quietly. Visibly. The kind where people stop calling and start whispering.

I called it a development year before that was fashionable. Volunteered. Worked without a title or a salary. Got humbled in ways a classroom never could. Treated that year like a product in beta — found the gaps, fixed them, shipped a better version.

Came out the other side not just recovered. Came out sharper. That skill — staying functional under pressure, making decisions with incomplete information — is not something you learn in a lecture. It's something you earn.

Chapter 06 · The Application
"I'm always ready for a war again, go down that road again — I'll ride again"
Pray For Me · The Weeknd

Everyone told me to stay. Work in your field. Stop chasing something nobody around you could picture.

I asked for one month to sort the funding. It took a year to get the loan approved. Applied to every university that mattered. Got rejected by all of them — first round. Then Hochschule Harz wrote back.

Small town in the Harz mountains. History, quiet, the kind of place that demands something of you. I said yes before I finished reading the letter. Conviction in a roadmap that hasn't proven itself yet. You back it anyway.

Chapter 07 · Germany
"I switch up my cup, I kill any pain"
Starboy · The Weeknd

Landing here felt like being a new product in an unfamiliar market — strong fundamentals, unclear positioning, everything to prove.

I missed home more than I admitted. Worked part-time, studied full-time, carried both without making it anyone else's problem. Found that every skill from Kolhapur — reading rooms, building teams, understanding what people need before they say it — translated perfectly.

The pain didn't disappear. I just got better at using it.

Chapter 08 · The Work
"But I won't cry for yesterday, there's an ordinary world somehow I have to find"
Ordinary World · Duran Duran

At Hochschule Harz I found the vocabulary for everything I'd been doing by instinct. Product strategy. Innovation management. Marketing with intent. Not new concepts — just the right words for decisions I'd already been making for years.

I built things. Assessed research. Designed innovation platforms. Presented to real stakeholders. Closed the gap between thinking and doing — because that gap is where most people live, and I refuse to.

Chapter 09 · November
"And it's hard to hold a candle in the cold November rain"
November Rain · Guns N' Roses

Nobody tells you that going after what you want is genuinely hard. Not exam-hard. Sustained, quiet, unglamorous hard. Tuesday in November in the Harz mountains when the sky is grey and you're not sure it's all going to work out.

I've held that candle through every version of this story. Kolhapur to Pune. Pune through failure. Failure onto a flight to Germany. Germany into whatever comes next — which I intend to be meaningful, in product management and marketing, at a company that builds things worth building.

The candle is still lit.

"Life is full of sweet mistakes and love's an honest one to make, time leaves no fruit on the tree"
John Mayer

I am not a finished product. Nobody worth hiring is.

What I am is someone who has moved through the world with intention — who has built teams, run events, crossed continents, learned from failure, and never stopped being curious about why people choose one thing over another.

That curiosity is the foundation of good marketing.
That execution is the foundation of good product management.
And that story — the football, the workshop, the lyrics, the rain — is why I show up differently.

See the work
Satyajeet Salokhe · 2025 Kolhapur → Pune → Wernigerode