AI Engineer
AI Engineer specialising in agentic systems, RAG applications and LLM workflows.
<coder>
Full-stack developer who writes clean, production-ready code — React, Node.js, Flask.
AI Engineer Intern @ ReadyForRole · Bangalore
> Agentic AI
I'm Bhaskar Pandit M N (ಭಾಸ್ಕರ್ ಪಂಡಿತ್) — an AI Engineer who turns disorder into working systems: agentic AI, LLM applications, RAG pipelines and full-stack products that hold up in the real world. Currently building production AI at ReadyForRole Pvt Ltd; previously teaching machines to catch faulty sensors at DRDO GTRE.
01 — About
I build things that think. Right now that means agentic AI systems, RAG applications and knowledge assistants at ReadyForRole Pvt Ltd — LLM workflows that don't just demo well, but survive real users in production.
Before that, I spent time inside GTRE (DRDO) — India's gas-turbine research establishment — training machine learning models to validate sensor data, identify faults and predict maintenance before things break. Aerospace research teaches you one thing fast: a model that's almost right is wrong.
As Co-Founder & Director of KLS Tech Solutions, I've shipped products end-to-end — like KLS Academy, a open-source internship platform — from architecture to deployment. Between the research rigor and the startup speed sits the work I like most: turning messy problems into systems people actually use.
Off the keyboard, I captain my college cricket team and have played state-level cricket for Karnataka. Leading on the pitch and designing systems share the same core skill: finding order in chaos.
02 — Experience
ReadyForRole Pvt Ltd
Building production AI end-to-end: agentic AI systems that reason and use tools, RAG applications over real knowledge bases, LLM workflows and knowledge assistants — shipped as full-stack products, not notebooks.
GTRE, DRDO · Bengaluru, Karnataka
Developed machine learning models for defence applications using TensorFlow and PyTorch at India's premier gas-turbine research establishment. Implemented AI-driven solutions for data analysis and predictive modelling in critical systems — where a model that's almost right is wrong.
M.S. Innovations · Bengaluru, Karnataka
Developed full-stack web applications using React.js and modern JavaScript frameworks. Built responsive user interfaces and implemented software development best practices — production work running alongside defence research.
KLS Tech Solutions · Bengaluru, Karnataka
Architected LokSuraksha, a full-stack citizen safety platform with a React Native mobile app and Node.js/Express API. Led development of Project Shakti, a women's safety app with SMS-based SOS alerts for offline emergency response.
Edunet Foundation · Bengaluru, Karnataka
Built responsive web applications using the MERN stack with secure authentication and API integration. Implemented security best practices and maintained 95% code coverage through testing procedures.
03 — Projects
Real systems with real users in mind — each one built problem-first.
Enterprise support teams drown in repetitive tickets, while single-shot chatbots lack the context and reasoning to resolve anything beyond FAQs.
A multi-agent customer support platform built with Agentic AI and Generative AI — Advanced RAG pipelines and Multi-Agent Systems via CrewAI that reason, retrieve and act on real support context.
AWS Bedrock, VectorDB and Embeddings integrated with PostgreSQL, orchestrated through LangChain and LangGraph — a scalable enterprise AI workflow, not a demo notebook.
Powers context-aware automated customer support: agents retrieve grounded answers from enterprise knowledge and hand off cleanly when a human is needed.
Turbofan test data is only as trustworthy as its sensors — a drifting or faulty channel can hide a real fault, or invent one that isn't there.
A production-ready predictive maintenance pipeline for GTRE/DRDO turbofan engines, achieving 96% fault classification accuracy across engine sensor streams.
An ensemble of XGBoost and Isolation Forest models behind a real-time Streamlit dashboard — live sensor validation, RUL prediction and anomaly detection in one view.
Turns raw engine test telemetry into actionable maintenance signals: bad channels flagged, remaining useful life estimated, anomalies surfaced before failure.
ML pipeline that predicts soil nitrogen (kg/ha) from 21-wavelength NIR spectral data using SVR and ANN models.
Full-stack citizen safety reporting ecosystem — mobile app, web dashboard and backend API in one platform.
AI-powered facial recognition system for contactless attendance management using a live camera feed.
More experiments and work-in-progress on github.com/BhaskPand →
04 — Skills
05 — Beyond Code
Played state-level cricket for Karnataka (2020) and awarded "Best Sportsman of the Year". Now captain of the APSCE college cricket team, with university-level match experience.
Coordinator of the ALGO Club at APSCE — organizing hackathons and running Git & GitHub sessions for students.
Team Head of the APSCE Music Society. I play 8 instruments — Mrudangam, Guitar, Flute, Ukulele, Kanjira, Djembe, Ghatam and Veena.
06 — FAQ
An AI Engineer and Full-Stack Developer based in Bengaluru, Karnataka, India. Currently a Full Stack Tech Intern at ReadyForRole Pvt Ltd, building AI enterprise solutions with Agentic AI, Generative AI, Advanced RAG and Multi-Agent Systems using CrewAI. Previously an AI/ML Intern at GTRE, DRDO, developing machine learning models for defence applications with TensorFlow and PyTorch. Also a Software Developer at KLS Tech Solutions, with a B.E. in Information Science Engineering from APS College of Engineering, Bengaluru.
Enterprise AI systems and production software: multi-agent workflows with CrewAI, Advanced RAG pipelines over vector databases, LangChain and LangGraph orchestration on AWS Bedrock, ML models for predictive maintenance and sensor validation, computer-vision systems, and the full-stack layer — React, React Native, Node.js, Flask — that turns models into products. Highlights include the Enterprise AI Support Platform, the Turbofan Sensor Validation pipeline (96% fault classification accuracy), and LokSuraksha (citizen safety platform).
Notable projects include the Enterprise AI Support Platform, a multi-agent customer support system built with CrewAI, LangChain, LangGraph, AWS Bedrock and Advanced RAG; Turbofan Engine Sensor Validation & Fault Identification, a predictive maintenance pipeline for GTRE/DRDO achieving 96% fault classification accuracy; a Soil Nitrogen Prediction System using NIR spectroscopy with SVR and ANN models; the LokSuraksha App, a full-stack citizen safety reporting ecosystem; and a Smart Attendance System using facial recognition with MediaPipe and OpenCV.
I work across the applied AI stack. At ReadyForRole Pvt Ltd I build AI enterprise solutions using Agentic AI and Generative AI with Advanced RAG and Multi-Agent Systems via CrewAI — working with AWS Bedrock, VectorDB, Embeddings, PostgreSQL, LangChain and LangGraph. At GTRE, DRDO I developed machine learning models for defence applications using TensorFlow and PyTorch, including a turbofan sensor validation pipeline with 96% fault classification accuracy. I also work in computer vision with OpenCV and MediaPipe, and classical ML with XGBoost, Scikit-learn, SVR and Isolation Forest.
Languages: JavaScript, TypeScript, Python, C++, HTML/CSS, SQL. Frontend: React.js, Next.js, React Native, Expo, Material-UI. Backend: Node.js, Express.js, Flask, RESTful APIs. Databases: MongoDB, SQLite, SQL, PostgreSQL, VectorDB. Agentic & Generative AI: Agentic AI, Generative AI, Advanced RAG, Multi-Agent Systems, CrewAI, AWS Bedrock, Embeddings, LangChain, LangGraph. ML & Data Science: XGBoost, Scikit-learn, TensorFlow, PyTorch, SVR, ANN, Isolation Forest, Pandas, NumPy. Tools: Docker, Git, SSH, Linux, OpenCV, MediaPipe, Streamlit, NIR Spectroscopy.
I write about AI engineering, agentic systems and applied machine learning on Medium. Code lives on GitHub, and my professional profile is on LinkedIn.
Every system tends toward disorder — the job of an engineer is to spend energy pushing back. Most of what I do is exactly that: taking chaotic inputs — messy sensor data, ambiguous prompts, half-formed product ideas — and turning them into systems that behave predictably. Chaos in, order out.
Yes — for AI engineering roles, agentic and generative AI product work, and interesting collaborations. I'm based in Bengaluru, Karnataka, India and work with teams locally and remotely. The fastest way to reach me is bhaskarpanditmn@gmail.com.
07 — Contact
An AI product to build, a system to untangle, or a role you think I'd be a fit for — my inbox is open.