Ayush Kothe
AI/ML Engineer
AI/ML Engineer specializing in Generative AI, RAG architectures, multi-agent systems, and production-ready machine learning applications. Experienced in building scalable AI solutions using Python, FastAPI, LangChain, Docker, and AWS.
About Me
I’m an AI/ML engineer with hands-on experience in Machine Learning and Generative AI applications across RAG, multi-agent systems and model deployment. I develop LLM-powered backend services, evaluate RAG pipelines, and deploy models using Python, FastAPI, Docker, AWS and CI/CD.
I’m a B.Tech student in Industrial & Production Engineering at SGSITS Indore (2024–2028), and I’ve interned as an AI/ML Engineer at M9Vends.
Core Expertise
RAG & Vector Search
Hybrid retrieval with Multi-Query and BM25 fused via Reciprocal Rank Fusion, with cited answers.
Multi-Agent Systems
LangGraph agents that combine technical and macroeconomic analysis into consensus signals.
Computer Vision
YOLOv8 and OpenCV pipelines for real-time activity detection on edge hardware.
LLM Evaluation
DeepEval suites for faithfulness, retrieval quality, prompt-injection and data-leakage defence.
ML Deployment
Containerised FastAPI services with Docker, GitHub Actions and cloud hosting.
Classical ML
Scikit-learn and XGBoost pipelines for classification, from data prep to prediction APIs.
Projects
Research Paper Assistant
RAG system for research-paper Q&A returning the top 5 cited chunks.
- LangChain, Groq LLM and BGE embeddings with hybrid retrieval
- 100-case DeepEval suite with a 0.70 pass threshold
- FastAPI, Redis and Streamlit on Render with LangSmith tracing
Agentic Crypto Trading Bot
Multi-agent trading system built with LangGraph.
- Binance OHLCV and CoinDesk RSS inputs
- SMC-based BOS, Order Block and FVG detection
- Groq LLaMA-3.3-70B with rule-based fallbacks and 1:2 risk-reward controls
Credit Risk Predictor
End-to-end credit risk classification pipeline: 87% accuracy, 84% F1-score.
- Python and Scikit-learn model
- FastAPI prediction API, Dockerised on Render
- GitHub Actions for testing, training and deployment
Experience
CTO & Founder · Connectify
- Architected and led end-to-end development of the platform (connectify.club), defining core system design and technology roadmap
- Built scalable backend services and API infrastructure to support reliable real-time operations
- Spearheaded product strategy, technical architecture, and engineering execution across the stack
AI/ML Engineer Intern · M9Vends, Indore
- Engineered ML classification models for real-time threat detection, reaching 92% accuracy with low-latency edge inference in mind
- Built a YOLOv8 and OpenCV pipeline, trained on custom-labelled data, to detect human activity and vandalism at 1 frame every 3 seconds
- Integrated vision-language models (Qwen2.5-VL and LLaMA-3) through a FastAPI backend for automated scene descriptions and threat alerts
B.Tech, Industrial & Production Engineering
Shri G. S. Institute of Technology and Science, Indore, MP
- Cleared 2 rounds of the SGSITS internal Smart India Hackathon 2026 selection
- Solved 200+ problems from Striver’s A2Z DSA Sheet
- Completed the Complete Data Science, ML, DL & NLP Bootcamp certification
Technical Skills
Generative AI
Machine Learning
Backend & APIs
Databases & Vector Stores
MLOps & Cloud
Programming & Tools
Let’s Build Something
Open to AI/ML internships, roles and collaborations on RAG, agents and deployed ML systems.