CV

AI Research Engineer at Aerobase specializing in Neural Combinatorial Optimization, Reinforcement Learning, and Robot Path Generation — training attention-based neural networks with modern policy-optimization methods (GRPO, PKPO, RSPO, PPO) under Pass@K / Max@K objectives to solve large-scale routing, scheduling, and planning problems. Strong foundation in machine learning, deep learning, and software engineering.

Work experience

2025 - present : AI Research Engineer, Aerobase, Sweden
Develop robot path generation and toolpath optimization for metal additive manufacturing (DED, LPBF, and WAAM), formulating deposition- and scan-path planning as a combinatorial optimization problem and solving it with neural combinatorial optimization and attention-based solvers to improve build quality, throughput, and material efficiency.
- Skills & Tools: Python, PyTorch, Neural Combinatorial Optimization, Attention Models & Transformers, Policy Optimization (GRPO, PKPO, RSPO, PPO), Pass@K / Max@K objectives, OR-Tools.
Apply reinforcement learning to defect minimization in additive manufacturing, training agents that reduce porosity, cracking, and residual stress using feedback from thermal and mechanical FEM, microstructure analysis, and melt-pool CFD simulations (simulation-in-the-loop optimization).
- Skills & Tools: Reinforcement Learning, FEM (thermal & mechanical), Melt-Pool CFD Simulation, Microstructure Analysis, Simulation-in-the-Loop Optimization.
2020 - 2025 : Doctoral Researcher, LTU, Luleå, Sweden
Developed a multi-agent AI framework for distributed control systems, integrating AI-driven decision-making and real-time task execution across industrial automation platforms. The system enables seamless agent collaboration, intelligent workload distribution, and optimized performance monitoring in decentralized environments.
- Skills & Tools: Multi-Agent Reinforcement Learning (MARL), MAS Frameworks, Distributed Computing, LangGraph.
Optimized LLM fine-tuning for generating IEC 61499 function blocks in industrial automation, enhancing accuracy and response times for AI-driven control system design. This involved iterative benchmarking and training of models to ensure optimal function block generation and seamless integration into automation workflows.
- Skills & Tools: PyTorch, LoRA (Low-Rank Adaptation), IEC 61499.
Developed an AI evaluation pipeline to test model responses for accuracy, factual consistency, and unintended biases, optimizing feedback analysis in an AI-driven workplace well-being platform. Tested various LLMs for benchmarking and evaluation using LangSmith to improve sentiment analysis and stress prediction from employee feedback surveys.
- Skills & Tools: PyTest, LangChain Evaluation, LangSmith, Benchmarking Frameworks, A/B Testing, AI Explainability Tools.
2019-2020: Full Stack developer, RCKR Software Pvt Ltd, Bengaluru, India
- Designed, developed, and deployed end-to-end machine learning solutions, including preprocessing, model training, evaluation, and deployment, using frameworks like TensorFlow, PyTorch, and Scikit-learn.
- Developed scalable and efficient full-stack applications integrating machine learning models, front-end interfaces, and back-end services using technologies such as Python, Next.Js, and DyanamoDB.
2017-2019: System Engineer, Tata Consultancy Services (TCS) , Kochi, India
- Integrated machine learning models into web and mobile applications by developing RESTful APIs and deploying models using cloud platforms like AWS and Azure, ensuring scalability and reliability.
- Designed and developed visually compelling dashboards and reports using tools like Power BI, or Python libraries (Matplotlib, Seaborn) to effectively communicate data findings to stakeholders.
2017: Data Analyst - Research Intern, Uvionics Tech India Pvt Ltd
- Designed, developed, and implemented machine learning models for disease prediction, utilizing algorithms such as logistic regression, random forest, support vector machine, and neural networks to analyze medical data and predict disease risks and outcomes.
- Evaluated and validated machine learning models using techniques such as cross-validation, ROC curve analysis, and confusion matrix to assess model performance and optimize models through hyperparameter tuning and ensemble methods to achieve robust and reliable predictions in real-world healthcare applications.

Education

B.S. in Electronics and Communication Engineering, SCMS, Kochi, MG University, India, 2014
M.S. in Computer Science and Engineering, NIT Trichy Campus, (IIIT), India, 2017
PhD in Industrial AI (Machine Learning & Formal Methods for Industrial Automation), LTU, Sweden, 2025

Skills

Neural Combinatorial Optimization: Attention Models, Transformers, Pointer Networks, Graph Neural Networks, constructive & improvement (local-search) policies, neural large-neighborhood search
Reinforcement Learning & Policy Optimization: Policy Gradients, Actor–Critic, REINFORCE, PPO, GRPO, PKPO, RSPO, Pass@K / Max@K (best-of-K) objectives
Combinatorial Problems: TSP, CVRP, job-shop scheduling, bin packing, assignment and routing
Optimization & Solvers: OR-Tools, LKH, Gurobi, metaheuristics, integer/linear programming
Machine Learning & Deep Learning: PyTorch, TensorFlow, Keras, Hugging Face Transformers, NumPy, scikit-learn
LLMs & Tools: OpenAI, Anthropic, DeepSeek, AWS Bedrock, Ollama, LangChain, LangSmith, LangGraph
Backend: Python, TypeScript
Storage: Postgres, DynamoDB, GraphDB, IPFS, S3
Containerization: Docker, Kubernetes
Cloud Computing: AWS
Software Development & Version Control: REST, GraphQL, Git, GitHub, Jenkins, GitLab CI