I'm Ankesh Babal — I ship LLM pipelines|
I'm an AI & data-focused engineer with a Master's in Analytics from Auckland University of Technology. My work sits where machine learning meets real operations — building models that are useful, pipelines that don't break, and systems that actually make it to production.
During my recent internship at Dealtable.ai, I built an end-to-end LLM system supporting venture capital investment decisions — processing 10K+ documents, designing a scoring and evaluation framework, and delivering outputs through a production-ready API.
I enjoy the full AI stack: from raw data ingestion on Azure and GCP, through retrieval-augmented generation, to the last mile where a model output becomes a decision. I'm now looking for a role where I can grow as an AI engineer and ship production AI systems.
Real-time VC investment decisions, served through a production API.
Isometric view · hover any layer to inspect
How I ship data from raw sources to dashboards · cloud-native
Languages, frameworks and platforms I use to move ideas from prototype to production.
A mix of AI systems, data pipelines, dashboards and ML projects. Source on GitHub.
End-to-end LLM system at Dealtable.ai. Ingested and structured thousands of VC documents, built a retrieval-augmented generation pipeline with Gemini API, designed a multi-criteria scoring framework, and shipped a sub-second decision API via FastAPI serving real-time investment insights.
Full transit analytics platform processing GTFS static & real-time trip updates. Built a 6-page Streamlit dashboard (headways, delays, reliability, heatmaps, route comparison, performance score). Designed cloud expansion on Azure Data Lake + Data Factory + Databricks.
Cloud-native data pipeline on Uber trip data — automated ingestion via Mage, transformation layer in Python, and analytics served through BigQuery on GCP. Full medallion architecture (bronze → silver → gold).
Classification models predicting obesity risk — compared SVM, Random Forest and XGBoost. SVM achieved 96.44% accuracy. Built a GUI for real-time input and prediction, targeted at healthcare screening workflows.
Interactive dashboard covering 5 business domains (Finance, Sales, Marketing, Supply Chain, Executive). MySQL source, snowflake modelling, DAX measures, DAX Studio performance tuning, and Power BI Service with auto-refresh & row-level security.
Expense Tracker (FastAPI + Streamlit), SQL Music Store analysis, AtliQ Hospitality dashboards, MILP route optimisation, and more. All source code public.
I don't just build models — I build systems that ship. From raw data ingestion to a production API, I own the full pipeline and make sure it actually works in the real world.
I've built production LLM pipelines over large-scale document sets, designed scoring frameworks for investment decisions, and shipped dashboards across finance, sales and supply chain. I work with real business data, not toy datasets.
AWS certified. Hands-on with Azure Data Factory, GCP BigQuery, Databricks. I don't just know cloud theory — I've built and deployed pipelines on all three major platforms.
Deep-diving into LangGraph, vLLM inference optimization, evaluation frameworks, and making LLM outputs reliable enough for real-money decisions.
Open to AI Engineer, Data Engineer and Data Analyst roles — full-time, contract or freelance.
babalankesh75@gmail.com