Hello, I'mPriyankesh
AI Engineer • Tech Enthusiast
About Me

I’m Priyankesh, and I’m currently pursuing my course in AI & ML under GGSIPU with a minor from IIT Ropar. My philosophy on learning is that if you want to learn something, you should build things instead of getting stuck in tutorial hell. Recently, I was selected in Google Summer of Code 2025 and I've also had some awesome runs winning national hackathons. My long-term goals include becoming a Data Scientist.If you have a fun idea or project in mind, feel free to reach out!
Work Experience
Open Source Developer
Extralit, Open Science Labs (GSoC 2025)June 2025 – October 2025- Developed an Enhanced AI OCR Extraction Pipeline using Marker, PyMuPDF4LLM, and OCRmyPDF, improving structured text extraction accuracy for scientific PDFs by 30%.
- Built an RQ-based document processing pipeline and integrated Modal for asynchronous execution, generating hierarchical Markdown outputs for LLM-based analysis, reducing processing latency by up to 5 minutes.
- Integrated a Weaviate vector database for semantic content retrieval and implemented a human-in-the-loop correction workflow, enhancing data reliability and annotation efficiency.
Data Analytics Intern
IBM SkillsBuild (CSRBOX)June 2024 – August 2024- Analyzed 70K+ financial transactions using Python and SQL to uncover fraud patterns and segment customer behavior, improving risk detection.
- Built customer segmentation models achieving 84% accuracy using ensemble techniques and feature engineering.
- Designed interactive Power BI dashboards that visualized customer segments and fraud trends, enabling business teams to make faster data-driven decisions and reducing report turnaround time by 25%.
Technical Skills
Programming Languages
Technologies & Proficiencies
Frameworks & Tools
Education
B.Tech in Artificial Intelligence & Machine Learning
University School of Automation and Robotics, GGSIPU - Delhi, IndiaNov. 2022 – May 2026Focusing on Artificial Intelligence, Machine Learning, and Data Structures.
Featured Projects
Deep Research AI Agent
Designed a multi-agent AI system with 3 specialized agents orchestrating end-to-end research workflows. Integrated Qwen-30B and ScrapeGraph AI, achieving 97% citation accuracy and 85% improvement in insight extraction across 20+ sources.
AI-Powered Web Scraper
Built automated data pipeline using Python & Selenium processing 1000+ URLs concurrently, reducing manual collection by 90%. Integrated GPT-4, Gemini, and Llama with Pydantic schemas achieving 95% accuracy in data extraction.
Samvaad - Discord AI Chatbot
Fine-tuned DialoGPT-medium (117M parameters) on 10k+ conversational datasets with 7-message context window, improving relevance by 25%. Deployed on Hugging Face Hub serving 200+ users with optimized memory management.
Recent Achievements
Amazon ML Summer School
Selected Participant
Top 3.5% selection among 10,000+ applicants.
Industrial Ideathon 2025
1st Runner Up
Awarded by Delhi CM for innovative industrial solution.
UST D3code Hackathon
National Winner
1st place among 8000+ teams, featured in Times of India.