Years of
Experience
I am Rahul
Machine Learning Engineer
Machine Learning Engineer with 7 years of experience designing and shipping end-to-end ML systems, recommendation engines, and NLP solutions. I work on production RAG, AI agents, and LLM applications — with the evaluation, grounding, and MLOps (Airflow, Docker, MLflow) that keep them running once the demo is over.
Available for freelance work · ~30 hrs/week · fixed-scope projects or monthly retainer · based in Bengaluru (IST), daily overlap with EU and US-East mornings
7
About Me
Most of my work now is getting teams past the prototype wall: the demo works, but it hallucinates, has no evaluation, and nobody is sure how to deploy or monitor it. I build the retrieval grounding, the citation enforcement, the agent orchestration, and the FastAPI/Docker/AWS layer that turns it into something you can put in front of users. Before GenAI I spent years on recommender systems and NLP pipelines in production, which is why I know when an LLM is the wrong tool.
Python
Machine Learning
Deep Learning
NLP
Generative AI
MLOps
Research
Published work behind the engineering — a retrieval method for small, domain-specific corpora, with the code open-sourced alongside the paper.
arXiv preprint
MODE: Mixture of Document Experts for RAG
A cluster-and-route retrieval method for small, domain-specific corpora that removes the need for an external vector database or a reranker. Competitive with standard pipelines on HotpotQA and SQuAD at substantially lower retrieval latency.
What I Build For Clients
Selected Work
Grounded RAG for a logistics SaaS
Client engagement. Their GenAI assistant answered shipment questions but no answer could be traced back to a source, so nobody trusted it in front of customers. Built the full pipeline — ingestion, chunking, embeddings, retrieval, citation enforcement — plus an eval set and red-teaming pass the team still runs. Result: 0.88 faithfulness and 92% citation coverage on the internal eval set.
#RAG #EVALUATION #CITATIONS
Multi-agent shipment-exception handling
Same client. Exceptions — delays, missing documents, address failures — were queued for humans one by one. Built a LangGraph agent with planning, tool use, routing, and shared state, escalating to a person only when it could not resolve a case. Result: ~70% of exception cases resolved without human escalation.
#AI-AGENT #LANGGRAPH #TOOL-USE #ORCHESTRATION
RAG-Citation
An open-source library combining Retrieval-Augmented Generation with automatic citation generation, attaching source references to generated content so every claim in an answer is verifiable. Published on PyPI: pip install rag-citation.
#SPACY #RAG
YouTube Script Writer
An open-source AI agent that generates tailored video scripts from a title, language, and tone. It supports formats from short clips to long-form content, collapsing research and drafting into one pass for creators.
#YOUTUBE #AI-AGENT #LANGCHAIN #LANGGRAPH #PROMPT-ENGINEERING
ytkit
Fetch subtitles from YouTube channels, playlists, and search results in a few lines of code — a corpus-building tool for RAG pipelines, AI applications, and content analysis.
#YOUTUBE
Recommendation Engine
An AI-powered recommender serving product-to-product and user-to-product suggestions across an e-commerce catalogue, combining collaborative filtering with content-based course-to-job mapping to lift discovery and conversion.
#Transformer #PyTorch #MATRIX-FACTORIZATION
My Experience
My Education
My Skills
GenAI
ML / NLP
MLOps
Backend
Let’s work together!
Tell me what you’re building, where it breaks today, and what “done” looks like. I’ll reply with a practical next step — not a sales pitch.
-
Email
rahulanand1103@gmail.com -
Phone
+91 98348 42796 -
Based
Bengaluru, India · IST (UTC+5:30) -
Availability
~30 hrs/week · fixed scope or retainer