AI/ML Engineer & Researcher | Generative AI, Agentic AI, RAG, LLMs, NLP & Backend Development

AI/ML Engineer skilled in Generative AI, LLMs, RAG, AI agents, NLP, and Python.
AI/ML experience at IIT Mandi and IIIT Dharwad, building intelligent applications and backend systems.
Co-authored an LLM benchmark with IBM Research collaborators

• Agentic AI
• C++
• C (Programmiersprache)
• JavaScript
• Large Language Models1 J.
• Node.Js
• Postgresql
• Python
• Pytorch
• Tensorflow

My work sits at the intersection of AI research, applied machine learning, Generative AI, and backend engineering. I enjoy solving practical problems by combining research, experimentation, retrieval systems, and software development.

**AI Engineering and Agentic Systems**

I have developed agentic workflows using LangGraph for multi-step query analysis, clarification, document retrieval, response generation, and reasoning. My work includes state persistence, retrieval-grounded responses, and integration with backend services.

**RAG Systems and Knowledge Retrieval**

I have built retrieval-augmented generation pipelines for legal documents and local file discovery. My experience includes PDF ingestion, text processing, embeddings, semantic search, vector indexing, OCR integration, incremental indexing, and source-grounded answer generation. I have worked with Milvus, Qdrant, and FAISS.

**AI/ML Research and LLM Evaluation**

I co-authored ORBIT, an evidence-grounded ITOps reasoning benchmark developed in collaboration with IIIT Dharwad and IBM Research. The work involved building a 3,178-instance benchmark, retrieval-grounded multi-LLM evaluation, agreement analysis with human annotators, and benchmarking five open-weight LLMs. The manuscript is under review at ICSE 2027.

I am also working on sequence-based EEG–EMG modeling, including a selective state-space modeling approach for reconstructing continuous EMG signals from EEG data. This research manuscript is under review at Elsevier.

**Machine Learning, NLP, and Computer Vision**

I developed a multi-label text classification system using the Reuters dataset with TF-IDF, character n-grams, and lightweight linguistic features. In my reported experiments, Macro F1 improved from 0.11 to 0.52, with Micro F1 reaching 0.83.

I have also trained YOLOv8 models for facial emotion recognition and developed real-time inference for image and video inputs. My broader experience includes model training, data preprocessing, experimentation, evaluation, and performance analysis.

**Multilingual Speech and Voice AI**

During my AI/ML internship at IIT Mandi, I worked on a multilingual voice pipeline integrating speech-to-text and text-to-speech components for Indian languages and English. The work involved IndicConformer/Whisper, IndicParler-TTS, CPU/GPU support, and integration with an agentic legal AI application.

**Backend and Full-Stack Development**

I build backend services using Python, FastAPI, Node.js, and Express. My experience includes REST APIs, JWT authentication, server-sent event (SSE) streaming, rate limiting, structured logging, PostgreSQL, MongoDB, and integration with AI pipelines.

I have also built YelpCamp, a full-stack web application featuring authentication, REST APIs, image uploads, geocoding, and server-side validation.

**What I Can Deliver**

I can help clients build RAG applications, AI agents, LLM-powered tools, semantic search systems, NLP and ML prototypes, model evaluation pipelines, data-processing workflows, and backend APIs. I value clear communication, reproducible experiments, maintainable code, and solutions grounded in measurable results.

I am open to suitable remote, part-time freelance projects involving AI/ML engineering, Generative AI, research implementation, data science, and backend development.

**Programming Languages:** Python, C, C++, JavaScript, SQL

**Generative AI and LLMs:** Retrieval-Augmented Generation (RAG), Agentic AI, LangGraph, LLM integration, prompt engineering, multi-step workflows, grounded generation, LLM benchmarking and evaluation

**Machine Learning and Deep Learning:** PyTorch, TensorFlow, scikit-learn, model training, model evaluation, classification, feature engineering, experimentation, performance analysis

**Natural Language Processing:** Multi-label text classification, TF-IDF, character n-grams, syntactic features, text preprocessing, embeddings, semantic similarity, information retrieval

**Vector Databases and Search:** Milvus, Qdrant, FAISS, vector indexing, similarity search, semantic retrieval, incremental indexing

**Backend Development:** FastAPI, Node.js, Express.js, REST APIs, JWT authentication, SSE streaming, rate limiting, structured logging

**Databases and Data Processing:** PostgreSQL, MongoDB, SQL, PDF ingestion, OCR integration, document processing, data cleaning and preprocessing

**Speech and Voice AI:** Whisper, IndicConformer, IndicParler-TTS, speech-to-text, text-to-speech, multilingual speech processing, CPU/GPU inference

**Computer Vision:** YOLOv8, facial emotion recognition, image and video inference, PyTorch

**Research and Evaluation:** Benchmark development, experimental design, multi-LLM evaluation, human-annotator agreement, statistical evaluation, sequence modeling, EEG–EMG modeling, selective state-space models

**Web Development and Tools:** HTML, CSS, JavaScript, Git, GitHub, VS Code, MongoDB, Cloudinary, MapTiler

**Research and Project Experience:**
- AI/ML Research Intern — IIT Mandi, LegalAgentic AI
- AI/ML Developer — Research and startup collaboration at IIIT Dharwad
- ORBIT ITOps benchmark — IIIT Dharwad × IBM Research, manuscript under review at ICSE 2027
- Sequence-based EEG–EMG modeling — manuscript under review at Elsevier
- Reuters multi-label text classification
- YOLOv8-based emotion recognition
- YelpCamp full-stack web application
- Hack Maze 2025 — Second place
- VidyutVega 2025 — Third place