Bhuvana K S
✉ work.bhuvanaks@gmail.com · 📍 Bengaluru, India · GitHub
Software engineer with experience in AI agents, graph databases, and backend systems. Skilled in building multi-database integrations and graph-based analytics across production and POC environments.
Work Experience
Software Engineer (Specialist Programmer)Dec 2024 – Present
Infosys Ltd., Bengaluru
- Fine-tuned small language models (SLMs) such as Phi-4-mini (3.8B) and Qwen3 using parameter-efficient techniques (LoRA/QLoRA) with Unsloth on ~1,500 curated domain-specific Q&A pairs across ~20 business domains, specialising responses for support-automation workflows.
- Integrated fine-tuned SLMs with Retrieval-Augmented Generation (RAG) over vector databases (ChromaDB, Pinecone) to keep domain knowledge current without full retraining, and deployed quantised (4-bit, ~4 GB) models locally via Ollama for private, low-cost inference with no cloud dependency.
- Built multi-database API integrations and backend services using FastAPI and WebSockets across 3+ heterogeneous database systems (document, graph, and relational).
- Built an end-to-end ML pipeline using Python and MongoDB for data ingestion, preprocessing, model training, and serialisation.
- Developed a human-in-the-loop validation interface using Gradio with automated retraining and champion model selection, cutting manual model-review effort by ~30%.
- Integrated AI agent workflows using MAF for intelligent automation across backend modules.
Software EngineerAug 2024 – Dec 2024
Knowledge Lens Pvt Ltd., Bengaluru
- Owned the Root Cause Analysis (RCA) module for manufacturing plant asset-malfunction analysis, modelling and storing graph data in Neo4j and JanusGraph.
- Built V1 of the company's internal backend SDK enabling engineers to visualise multiple graph layouts through a single unified function call, cutting graph-visualisation boilerplate by ~50%.
Software Engineer InternFeb 2024 – Aug 2024
Knowledge Lens Pvt Ltd., Bengaluru
- Handled WebSocket API for the FTDM chatbot feature; optimised Cypher and Gremlin queries in Neo4j and JanusGraph achieving 40% faster query performance and 25% reduction in data retrieval time.
- Built a Gremlin graph visualiser using vis.js and Python; led the Kafka version upgrade for Jublient's client module.
Data Science InternNov 2023 – Dec 2023
Technofly Pvt Ltd., Bengaluru
- Worked in Python with NumPy, Pandas, Tkinter and TensorFlow; explored machine learning and deep learning algorithms including YOLO and Keras classification.
- Built a mini-project on the Google Speech API that converts voice to text and back to an automated voice.
Projects
- graph-viz: JanusGraph Gremlin traversal paths as interactive graphs with vis.js.
- FinVoice: voice-first money management with group bill splitting and on-chain settlement.
- graph-layouts: Python SDK that loads graph layouts by name.
- Travel Book: a digital scrapbook for travel memories.
- More on github.com/bhuvanaks02, or the full list.
Skills
- Languages
- Python, C++, SQL, Gremlin, Cypher
- AI & ML
- Object Detection, MAF, SLM Fine-tuning, Unsloth, Hugging Face, Ollama, RAG
- Backend
- FastAPI, WebSockets
- Databases
- PostgreSQL, MongoDB, Neo4j, JanusGraph, ChromaDB, Pinecone
- Tools
- Git, vis.js, Gradio
Education
Bachelor of Engineering in Computer Science2020 – 2024
Global Academy of Technology, Bengaluru · CGPA: 9.17
Publications
- ML-Based Graduate Admission Prediction. Published at the IEEE International Conference on Intelligent Systems, Communication and Applications (CIISCA 2023).
Achievements
- Hackathon Judge: Invited as a judge at 3 college hackathons across Global Academy of Technology and Dayananda Sagar University (twice).
- Hackathon Winner, Evolve 2024: 1st place in the office hackathon at Knowledge Lens.
- 1st Place, College Hackathon 2021: Built a complete front-end website using HTML and CSS.