Computer Science Engineer · Data Scientist · Generative AI · MLOps · LLMOps 📧 [email protected] · LinkedIn · GitHub
Hi — I’m Vijay. I turn messy data into clear decisions and prototypes into reliable ML/AI products. I enjoy building end-to-end systems (data collection → models → production) that actually make people’s lives easier — for example, tools that summarize long videos, automate job search workflows, or score resumes against real job descriptions. I combine practical engineering with a focus on evaluation and user impact.
🔎 Quick highlights
- Built a video transcription + summarization agent that converts long recordings into concise, context-aware summaries for learners and creators.
- Built an automated job-search & career assistant that parses resumes, fetches job listings, suggests improvements, and runs practice interviews — designed to reduce search friction and improve interview readiness.
- Intern experience in Generative AI and Data Science — shipped models and production-ready pipelines, improved chatbot UX via prompt engineering and model tuning.
- Focus areas: GenAI, RAG, MLOps, model evaluation, and production reliability.
What I build
- End-to-end ML/DL systems (research → prototype → production).
- LLM-powered tools: retrieval-augmented generation (RAG), embeddings + vector DBs, prompt engineering and tuning.
- Robust APIs and interfaces (FastAPI, Streamlit, React) for real users.
- Reliable pipelines and monitoring (CI/CD, tests, model & data validation).
Featured tech (Modern & Practical)
- Languages: Python, SQL, Bash
- ML/DL: PyTorch, TensorFlow, Hugging Face Transformers, Diffusers, Whisper
- LLM / GenAI: LangChain, Llama / Llama2, Bloom, Mistral, Vicuna, instruction tuning, LoRA, PEFT, RLHF techniques
- Retrieval & vector DBs: FAISS, Milvus, Weaviate, Pinecone
- MLOps / infra: MLflow, Weights & Biases, Seldon / BentoML, Triton, ONNX, Docker, Kubernetes
- Orchestration & infra: Airflow, Prefect, Ray, Dask, AWS (EC2, S3), Databricks, Snowflake
- Observability & testing: Prometheus, Grafana, Great Expectations, Evidently
- Data & visualization: Pandas, NumPy, Spark, Plotly, Streamlit, Dash, Tableau
- Deployment & optimization: CUDA, TensorRT, quantization (bitsandbytes), model compression, CI/CD
(If you want a full list of libraries I use in projects, ask — I keep a living list.)
Featured projects (click to expand)
AI-agent — Transcribe & Summarize Videos
- What: End-to-end agent that transcribes long videos (Whisper) and generates short, context-aware summaries and chaptering.
- Why: Faster content consumption, better accessibility and searchable clips for learners and creators.
- Tech: Whisper, Hugging Face Transformers, LangChain, FastAPI, Streamlit, FFmpeg, Docker.
- Result: Prototype for reducing time-to-insight for long-form content — easy to deploy as a lightweight service.
Automated Job Search & Career Assistant
- What: Agent that parses resumes, scrapes or calls job portals, ranks jobs by fit, suggests targeted resume edits, and runs mock interview prompts.
- Why: Saves time and improves candidate preparedness with concrete action items.
- Tech: Python, BeautifulSoup/Requests, job APIs, NLP embeddings, Transformers, simple web UI.
- Result: A workflow that automates repetitive job-search tasks and surfaces high-impact resume changes.
LLMOps Playground (experimental)
- What: Small platform for experimenting with RAG patterns, vector stores, prompt templates, and deployment monitoring.
- Why: Fast iteration on prompt strategies, retrieval layers, and performance metrics before productionizing.
- Tech: FastAPI, FAISS/Milvus, Docker, W&B / Prometheus for metrics.
And many more
Open-source toolkit — Resume Analyzer (idea → prototype)
- What: Pipeline that scores resumes vs job descriptions and provides explainable edits and examples.
- Why: Reusable, practical tool for applicants and small recruiters.
- Tech: NLP embeddings, cosine-similarity, attention to explainability and UX.
More projects & demos
- Tiny apps: demo dashboards, model cards, and evaluation notebooks.
- Automations: scheduled ETL, daily data pulls, simple monitoring jobs.
- Research protos: prompt engineering experiments, low-cost fine-tuning, small-scale RLHF sketches.
End-to-end data analytics skills including: - Data cleaning and preprocessing - Exploratory Data Analysis (EDA) - Data visualization best practices - Business Intelligence (BI) reporting - Dashboard development for insights
Experience (select)
- Generative AI Intern — improved LLM chatbot UX with targeted prompt engineering and tuning; prototyped RAG workflows and evaluation pipelines.
- Data Science Intern — built ML models and deep learning pipelines, delivered reproducible training code and deployment blueprints.
- LLMOps Intern — contributed to VertexOps LLMOps platform for GenAI deployment: implemented end-to-end model deployment pipelines (CI/CD), prompt engineering and RAG integration, monitoring and evaluation tooling, and delivered deployment blueprints and documentation.
How I work
- Start with the user problem and measurable success criteria.
- Prototype fast, evaluate rigorously, then harden the best solution for production.
- Prioritize reproducibility, tests, and clear monitoring — small measurable wins beat big untested ideas.
- Prefer modular architectures: decoupled ingestion, transformation, model, and serving layers.
Open to
- Full-time roles in ML/AI engineering, Data Science, or GenAI product engineering.
- Contract work: prototyping LLM products, RAG systems, MLOps pipelines.
- Collaborations: open-source tooling, research-to-prod prototypes, or product-minded AI features.
Contact
- Email: [email protected]
- LinkedIn: https://www.linkedin.com/in/vijaykalore
- GitHub: https://github.com/vijaykalore
If you want something built that actually ships and helps people, let's talk. I enjoy turning curiosity into production-ready systems.
— Vijay
Last updated: September 2025
