Career & Background

I'm an engineer and consultant with a decade of experience in data, analytics, and software — delivered across retail, banking, insurance, FMCG, and manufacturing. My career has been shaped more by exposure to real operational problems than by chasing the latest technology trends. This post is a factual catalog of education, roles, industries, and the approach that emerged from that experience.

Education

B.Tech — Amrita School of Engineering Graduated January 2015. Provided the foundational engineering and analytical toolkit that has underpinned every role since.

Career Timeline

May 2015 – July 2018 — Data Scientist, Mu Sigma Bengaluru, KA. My first three years in the industry were a wide-angle lens: insurance, retail, food services, banking, and government transport. Projects ranged from predictive churn and retention models for a major Asia-Pacific insurance client, to a forecasting system for a food services company improving logistics efficiency through sales predictions, to generating synthetic test data for a Swiss bank's offshore development. Also built the foundational R-based ML platform for a global analytics department and ran presales across multiple verticals.

July 2018 – June 2019 — Associate, TheMathCompany Bengaluru, KA. Embedded onsite at AB InBev within their People Analytics team. Built a global attrition risk framework processing data for ~6,000 employees worldwide. Simultaneously led technical hiring panels and trained incoming data science talent at the firm.

June 2019 – May 2020 — Data Science Specialist, Bain & Company Bengaluru, KA. Strategy consulting meets internal IP development. Built a Supplier Negotiation Tool for the Retail practice, deployed for clients including Metro China. The role sharpened my ability to frame technical work in terms of business outcome rather than model performance.

May 2020 – May 2022 — Lead Data Scientist, Tesco Bengaluru, KA. Designed and built an end-to-end, in-house A/B testing and experimentation platform from scratch — covering everything from physical store layout changes to digital product launches. The system was built on the Hadoop ecosystem with a Dash web front-end, deployed across custom VMs and load balancers on their internal cloud infrastructure. Two years of full ownership of a production-grade analytics product.

May 2022 – March 2024 — Global Analytics Manager, AB InBev Bengaluru, KA. Leading data scientist for the global Supply function. Acted as the primary liaison between six global zones and central HQ on brewing quality, efficiency, AI/ML platform development, and operational analytics. Managed a portfolio of approximately $1M USD CAPEX/OPEX delivering $2–3M USD in operational benefits.

March 2024 – January 2025 — Senior Manager, AB InBev Bengaluru, KA. Spearheaded SODA AI — an in-house Advanced Process Control framework for end-to-end brewery automation. Architected the system to stream real-time data from edge devices on the brewery floor to the cloud, run predictive ML models, and actuate physical equipment controls to optimize brewing and packaging output. Stack: FastAPI, Azure, Databricks/Unity Catalog, Medallion Architecture, Docker/Portainer, CI/CD.

2025–Present — Independent Stepped back from full-time corporate engagement. Now selective about projects, focused on game development, mobile apps, and applied tooling as an independent solo developer through ItamiForge.

Industry Exposure

A decade of work cut across:

• Insurance: Churn/retention modeling, synthetic data generation, presales.
• Retail: Experimentation platforms, supplier negotiation tooling, forecasting, store operations analytics.
• Banking: Synthetic test data for offshore development.
• People Analytics / HR: Global attrition modeling at scale.
• FMCG / Brewing: End-to-end supply analytics, advanced process control, brewery automation.
• Consulting: Strategy delivery at Bain & Company; onsite embedded consulting at AB InBev via TheMathCompany.

The pattern across industries is consistent: operational problems are more often stuck on problem definition and data quality than on model sophistication.

Technical Stack

Languages: Python, R, SQL, TypeScript / JS, Bash / Shell

Data Science & AI: XGBoost, predictive modeling, forecasting, A/B testing, generative AI, applied ML

Databases & Big Data: PostgreSQL, Snowflake, Redis, Neo4j, Hadoop / Hive, MySQL, SQLite

Cloud & Infrastructure: Azure (primary), AWS, GCP, on-premises systems, Docker, CI/CD, Linux / VMs

Web: React / Next.js, Astro, Dash / Flask

Games & Apps: Bevy, Ebitengine, Pygame, Expo

Approach

A few things I've learned to hold consistently:

Clarity over complexity. The best solution is usually the simplest one that solves the actual problem. Generalized platforms and frontier models are tempting; the right tool for your specific constraint is rarer and more durable.

Operational grounding matters. Brewery automation is different from insurance modeling not because the math is different, but because the physics, the organizational risk tolerance, and the feedback loops are different. Ignoring domain constraints produces systems that work in demos and fail in production.

Ownership by design. I build for the teams that inherit the work. Minimal licensing lock-in, observable pipelines, understandable architecture. The goal is that the client's engineers can extend and maintain without me.

Now

I'm now building independently through ItamiForge — games, mobile apps, and developer tooling. This site documents those projects, the tools, and the writing that came from both sides of the work: the decade of applied engineering and what came after.