Rajarshi KarmakarSoftware Engineer & AI Researcher

Building intelligent systems that scale.

Specializing in production-ready AI systems.

About Me

I am an Software Engineer passionate about developing robust, scalable and impactful intelligent AI systems. My expertise lies at the intersection of deep learning research and production-grade software engineering, allowing me to build solutions that are not only state-of-the art but also practically viable in real-world.

My AI philosophy centers on creating systems that are transparent, interpretable, and aligned with human values. I am particularly interested in AI systems that can be interated to businesses and existing software applications and how autonomous systems can collaborate with humans to solve complex problems.

Currently Exploring

Building and experimenting with LLM models
Advanced agentic workflows with LangGraph and AutoGen
Optimizing RAG pipelines with performance metrices

I am open to relocating and remote opportunities that allow me to contribute to challenging AI projects.

Key Strengths

Retrieval-Augmented Generation (RAG)
Agentic AI Systems
Transformers & Attention Mechanisms
Backend Engineering
Python for AI/ML & software development
Building and deploying AI models from scratch
System design

Technical Skills

Core AI/ML

Transformers

CNNs

RAG Pipelines

Deep Learning

NLP

Agentic Systems

Frameworks

PyTorch

TensorFlow

scikit-learn

Hugging Face

Autogen

Tooling

Git

Docker

Kubernetes

Backend

FastAPI

Node.js

PostgreSQL

Kafka

Infra/DevOps

CI/CD

GCP

AWS

Terraform

Teamcity

Ansible

Vector DBs / LLMs

Pinecone

Weaviate

LangGraph

LangChain

Projects & Case Studies

SuperSegmentation: Novel architecture for Keypoint matching

A high performance Keypoint detection and matching model with the semantic context for SLAM in robotics, achieving state-of-the-art benchmark for cityscape.

Role: AI Researcher at Valeo

Pytorch

Tensorboard

Git

CUDA

Live Demo GitHub

SafeSpaceAI: AI platform for Clinical Therapists

A Gemini AI based platform to analyze patients mood in real time, an AI Chatbot for giving therapy to patients and a whole platform to monitor patients' mood.

Role:

ReactJs

Typescript

Gemini

NodeJs

Live Demo

LangChain/LangGraph: Open-Source Contributor

Contributed significantly to a popular open-source Agentic AI framework, focusing on performance optimization and new feature implementation.

Role: OSS Contributor

Python

Git

Github Actions

GitHub

SelfDriving Car Simulation Agent

Self-Driving Car Simulation using Reinforcement Learning aims to create an environment where an agent learns to drive a car autonomously

Role: OSS Contributor

Python

Git

Pytorch

Reinforcement Leanring

Pygame

Live Demo GitHub

Experience Timeline

AI Researcher (Internship)

Valeo Vision Systems

Feb 2025 - Aug 2025

Designed and implemented a novel computer vision model (“SuperSegmentation”) that combines image understanding with precise feature detection, supporting self-driving car navigation (SLAM).
Conducted a comprehensive review of 50+ research papers to identify gaps in existing methods, discovering a large opportunity for improving consistency and precision in visual mapping.

Laboratory Instructor (Contract)

University of Limerick

Sep 2024 - Nov 2024

Supervised and guided students in Operating Systems labs, ensuring they understood both theory and practical applications.
Delivered lessons on core OS concepts such as process management, memory handling, and file systems.

Software Engineer (Full-time)

Ciena

Apr 2022 - Aug 2024

Contributed to the development of network software powering Ciena’s optical platforms, helping customers handle massive internet traffic growth.
Designed and improved cloud-ready microservices, ensuring reliable communication between network devices and management systems.

Software Engineer (Full-time)

HCL Technologies

Feb 2021 - Apr 2022

Supported large-scale Linux system deployments, reviewing code changes and applying patch releases to keep systems secure and reliable.
Built monitoring dashboards using Grafana and Prometheus, giving teams real-time visibility into system health (CPU, RAM, database performance).