Harsh Yadav

Product Engineer

Delhi, India2 yrs 7 mos experience

Most Likely To SwitchHighly Stable

Key Highlights

Strong foundation in AI and ML with hands-on research experience.
Proven leadership in organizing tech events and hackathons.
Skilled in NLP and Computer Vision projects.

Stackforce AI infers this person is a Data Science and AI enthusiast with a focus on research and event management.

Contact

Skills

Core Skills

NlpResearch SkillsComputer VisionEvent Management

Other Skills

Applied Machine LearningArtificial Intelligence (AI)Data CleaningData ScienceData VisualizationDeep LearningDeepfake DetectionEditorial ProductionEvent PlanningExploratory Data AnalysisFunctionalityGoogle Cloud Platform (GCP)GraphsLLMsLeadership

About

Master's student at Department of Computer Science, Faculty of Mathematical Sciences, University of Delhi, India. Completed Bachelor of Science in Physical Sciences from Maharaja Agrasen College, University of Delhi, India in 2023. Apart from the curriculum, I am a passionate reader & learner of ML& DL. Learning the advance mathematical concepts of Probability & Statistics, Linear Algebra and Calculus. Research experiences: AI In Healthcare Computer Vision NLP

Experience

Ashoka university

Pre-Doctoral Research Workshop

Jan 2025 – Jan 2025 · 0 mo · Haryana, India · On-site

Department of csis bits pilani goa campus

Fifth Indian Symposium on Machine Learning (IndoML)

Dec 2024 – Dec 2024 · 0 mo · Goa, India

Indian institute of technology, delhi

Learning on Graphs New Delhi Meetup 2024

Nov 2024 – Nov 2024 · 0 mo · New Delhi, Delhi, India

Jointly organised by Mastercard AI Garage and Yardi School of AI, IIT Delhi

Graphs

Department of computer science, university of delhi

5 roles

Student Researcher

Promoted

Jul 2024 – Jul 2025 · 1 yr

Triplets extraction for Knowledge Graph Creation of Legal Documents (NLP):
◦ Collaborated on extracting triplets from legal documents for Knowledge Graph (KG) creation using LLMs such as LLama and Mistral.
◦ Implemented text preprocessing and abbreviation handling with NLTK and SPACY for accurate data representation.
◦ Experimented a pipeline based on recent works (EDC) for triplet extraction, achieving notable performance with LLama 3.1-8b.
◦ Currently optimizing pipeline performance and expanding triplet extraction to 1000 additional judgments.
◦ Supervisors: Dr. Vasudha Bhatnagar (Senior Professor, DUCS) and Dr. Vikas (Assistant Professor, DUCS).

Peer ReviewsExploratory Data AnalysisFunctionalityNLPResearch Skills

Event Organizer

Apr 2024 – Apr 2024 · 0 mo

Organized HackDUCS, a hybrid inter-university hackathon as part of the 20th Annual TechFest - Sankalan 2024.
Hackathon Tracks:
1. AI for India🧑‍💻 - Innovate for India and present unique ideas that inspire the nation’s AI journey. Focus on empowering the Indian society with toolkits to promote citizen participation in vernacular languages.
2. Human-Centred Design 🤖 - Develop and design solutions that address complex and everyday socio-technical problems with tools like accessible design, GenAI and XR.
3. Sustainable Technology 🌱 - With focus on low-carbon footprint, come up with solutions to enable and monitor sustainable development at the grassroot level.
4. Sponsor Track: Polygon 🔗 - Build solutions on Polygon that democratize Web3.
5. Sponsor Track: ETHIndia 🔗 - Build an innovative project on the Ethereum blockchain.

Event PlanningTeam LeadershipEvent Management

Student Researcher

Promoted

Mar 2024 – Jun 2024 · 3 mos

Research problem statement: Given an A-V sample containing a single speaker, the task is to identify if the video is a deepfake or real. If fake, find out the timestamps [start, end] in which the manipulation is done.
Tasks Accomplished:
Set up the server infrastructure from scratch to facilitate the research work.
Conducted comprehensive literature review and analysis of existing architectures.
Reproduced results of two state-of-the-art research papers using NVIDIA A100 GPU.
Contributed to the UMMAFormer’s GitHub repository, Updated required packages in the README file.
Corrected the file structure flow chart as provided by the author.
Supervisor and Co-Supervisor : Dr. Bharti (Assistant Professor, DUCS) and Dr. Vasudha Bhatnagar
(Senior Professor, DUCS

Research SkillsTeam LeadershipWell CompletionComputer Vision