AI/ML Engineer · LLM Systems · RAG Pipelines · FastAPI Backend
Building production AI systems that ship — not just notebooks that demo
I'm an AI engineer currently working on agritech AI systems at Kamal-Paterson Ltd (KPCloud) — crop health monitoring, RAG-based farming advisory, and computer vision features in production.
Before that: multi-agent systems, semantic QA, RAG pipelines, and backend APIs — mostly Python + FastAPI + LangChain/LangGraph. I work on the full loop from model to deployed API.
Kamal-Paterson Ltd · Agritech · Python, OpenAI GPT-4o, FastAPI
Designed the retrieval pipeline for an AI farming advisor that gives crop recommendations based on weather forecasts, soil conditions, and time-series trends. Built the RAG layer using curated crop data files + external APIs (Open-Meteo, geocoding). Also contributed the bilingual Bengali/English voice interface using GPT-4o.
Kamal-Paterson Ltd · Python, FastAPI, Flutter, React.js, PostgreSQL/PostGIS
Designed the database schema and FastAPI backend for a field issue reporting system used by tea estate managers. Built the geospatial data layer (PostGIS), offline-first Flutter mobile app, and React dashboard. Integrated weather data and a treatment tagging system (pesticide, fungicide, fertilizer).
SRS Studio — Multi-agent SRS generator
Multi-agent pipeline (planner → architect → developer → QA → reviewer) that converts a one-line software requirement into a full structured SRS document. Built with LangGraph, FastAPI, PostgreSQL. Supports ERD generation, run tracking, and Markdown export.
Python LangGraph FastAPI PostgreSQL Docker Multi-agent
ResumeRAG — Bilingual RAG resume analyzer
Upload any PDF resume → ask natural-language questions about skills, experience, education, certifications. Bilingual (Bengali + English). Source-backed answers, no hallucinations. Built with FAISS, Groq API, FastAPI.
Python RAG FAISS Groq API FastAPI NLP
Question Answering System — Semantic QA with BERT
Intent-aware QA system using BERT sentence transformers + FAISS. Understands paraphrased questions, returns answers with confidence scores. Modern UI.
Python BERT FAISS Flask Semantic Search
SkyScout AI — AI weather companion
Location-aware weather assistant that gives AI-powered activity recommendations based on live conditions. Uses Open-Meteo + geocoding + Groq API + FastAPI.
Python FastAPI Groq API Open-Meteo Geocoding
Smart Campaigner — RFM customer segmentation
Segments customers into Bronze/Silver/Gold/Platinum tiers using RFM analysis on transaction data. Sends personalized email campaigns per segment. Built with Flask + K-Means clustering.
Python Flask K-Means RFM Analysis Jupyter
Gesture Controller for Hill Climb Racing — Computer vision game control
Real-time hand gesture → keyboard control mapping using OpenCV and MediaPipe. Play games touchlessly. Responsive gesture-to-key pipeline.
Python OpenCV MediaPipe PyAutoGUI Computer Vision
| Area | Tools |
|---|---|
| Languages | Python, C/C++ |
| Backend | FastAPI, Flask, REST APIs, JWT Auth |
| AI / LLM | LangChain, LangGraph, CrewAI, Haystack, HuggingFace, OpenAI API, Groq API |
| ML / DL | PyTorch, TensorFlow/Keras, Scikit-learn, Transformers |
| Retrieval | FAISS, Vector DBs, RAG pipelines, semantic search |
| Databases | PostgreSQL, PostGIS, MySQL |
| Infra / Tools | Docker, Git, MLflow, Roboflow, n8n, Make.com, Zapier |
| Data | NumPy, Pandas, Matplotlib, Seaborn, Plotly |
An Interpretable ML Framework for Predicting Structural Heart Disease from ECG Using LLM-Powered Insights ML + LLM integrated framework for ECG-based structural heart disease prediction. ROC-AUC ~0.987 with explainable, clinically actionable outputs.
- 🏆 Top 50 — NASA Space Apps Challenge 2024 — Led ExoQuest, an exoplanet discovery and survival training platform
- 🏆 Finalist — Take-off Programming Contest 2022
- 💻 300+ competitive programming problems solved (Beecrowd, Codeforces, CodeChef)
- 🤝 Junior Community Manager, Machine Learning Bangladesh — organizing sessions, workshops, and mentoring beginners
B.Sc. Software Engineering (Major: Data Science) · Daffodil International University · 2022–2026 CGPA: 3.62 / 4.00