Samyak Nahar — AI Engineer

AI Engineer in Pune, India. I build production agentic systems, RAG and ML pipelines end-to-end. Currently driving the technical direction of Retail Agent at Propheus. Open to AI Engineer and Forward-Deployed AI Engineer roles. Contact: samyaknahar004@gmail.com · GitHub · LinkedIn · Resume

Experience

• Propheus — AI-ML Engineer (Jul 2025–present): Driving Retail Agent, an agentic system giving retailers a 30-day forward view of events, competitor promotions and weather. OpenAI and Gemini agents with custom LlamaIndex tools and ChromaDB retrieval classify store-relevant signals; a deterministic model computes exact USD revenue impact. Average +8% per-store revenue lift for a leading US retail brand.
• Syngenta — AI-ML Intern (Jan–Jul 2025): Sales forecasting for 40+ products across 4 countries at 5–14% MAPE on AWS SageMaker; XGBoost demand modeling; NL-to-SQL agents on AWS Bedrock serving 10,000+ non-technical users over a 23GB database.
• EduPlus — GenAI Intern (Aug–Dec 2024): Llama 3 Q&A chatbot (FastAPI, React) serving 1,500+ student inquiries a year, deployed on SageMaker with CI/CD to EC2.
• Millennium Semiconductors — AI-ML Intern (Feb–Jul 2024): Cross-brand electronic-component matching engine over data scraped from 50+ sources — deterministic spec comparison with embedding-based semantic fallback. Commended org-wide.

Projects

• NyayaBodh — GenAI legal research engine, 96% NER accuracy; Smart India Hackathon 2024 Grand Finale runner-up.
• AI Video Navigation — Whisper + LLaMA over 300+ hours of training video.
• Transformer From Scratch — pure PyTorch: multi-head attention, positional encoding, custom BPE tokenizer.

Awards

• Runner-up — Smart India Hackathon 2024 Grand Finale (380 teams)
• Runner-up — DataHacks III Hackathon
• Winner — Vortexa Tech Hackathon (250+ participants)

Skills

Python, LlamaIndex, RAG, LLM agents and tool use, OpenAI, Gemini, AWS Bedrock, ChromaDB, pgvector, PyTorch, XGBoost, HuggingFace, SageMaker, Docker, FastAPI, React, PostgreSQL.

Hiring page · Plain-text profile for AI assistants (llms.txt)