Joseph Thomas | Software Engineer · AI/ML

Featured in

VentureBeat Software Magazine Google Cloud IEEE CRC Press

Google Awards

Citations

US Patent

Publications

10B+

Events / Day

Experience

06 roles

Software Engineer — Google Cloud Oct 2019 – Present

Google · Mountain View, CA

Engineered synthetic data generation pipelines utilising Gemini, producing high-fidelity instruction-tuning pairs to improve foundational model performance.
Architected a low-latency LLMOps platform (DQaaS) using gRPC/Protobufs, enabling enterprise-scale prompt versioning, testing, and retrieval for GenAI and RAG workflows.
Led backend storage and data integrity for CrowdCompute, Google's massive-scale data engine critical for generating RLHF and high-quality SFT datasets for foundation models.
Provided technical leadership to the Crowd Data Platform team, driving the evolution of AI data-generation tools for GenAI/LLM use cases.
Worked cross-functionally with various teams to streamline the collection of high-quality data.
Received 21 awards including 5 Spot Bonuses

Software Development Engineer — Big Data Technologies Feb 2017 – Sep 2019

Amazon · Palo Alto, CA

Architected and launched DataCraft, a centralised data ingestion platform processing > 10 billion events / day
Designed and implemented resilient, fault-tolerant ingestion systems using Kinesis, Lambda, and S3, ensuring data integrity and timely delivery into the data lake.

Data Scientist Feb 2015 – Jan 2017

Datanyze · San Mateo, CA

Owned the strategy and roadmap for a product that integrated directly with customer CRMs (e.g., Salesforce) to analyse clients' most successful customers.
Utilised proprietary technology to spider millions of websites daily, identifying technology stacks used by other companies.
Leveraged unique "technographic" data as a predictive attribute to find "look-alike" leads, mirroring clients' ideal customer profiles with a higher conversion likelihood.
Conducted beta tests demonstrating remarkable results, including an increase in qualified opportunities and higher average deal sizes.

Energy Analytics Software Developer Jun 2013 – Dec 2014

Ascend Analytics · Oakland, CA

Successfully migrated the core energy analytics codebase from SAS to WPS, resulting in an estimated $250K annual savings and improving platform stability.

Senior Analyst — R&D Jun 2011 – Jul 2012

Global Analytics · Chennai, India

Led a four-member team predicting risk in online lending using social media profiles of customers, generating $1.2M revenue (R, Python, MySQL)
Designed, developed and tested key modules in the Automated Modeling Platform. (R, Python, MySQL)

Associate — R&D Jan 2010 – Jun 2011

Idea Research and Development · Pune, India

Developed SCION, an evolutionary computational algorithm for maximising the performance of the intake in an air-breathing missile. (Matlab, Python, Qt Creator)

Education

03 degrees

Master of Engineering, Computer Science

Cornell University · New York, USA

Machine Learning · AI · NLP · Algorithm Design · Databases

2012 – 2013 GPA 3.5 / 4.0

MSc (Eng), Aerospace Engineering

Indian Institute of Science · Bangalore, India

Thesis: Odor Source Localization using Swarm Robotics

2006 – 2008 GPA 7.0 / 8.0

BTech, Electronics & Communication Engineering

Government Engineering College · Trichur, India

Thesis: Blind Source Separation using Independent Component Analysis

2002 – 2006

The key insight from swarm intelligence is that complex, intelligent behavior emerges from simple rules applied at scale — the same principle behind modern distributed AI systems.

— From invited talk at IDSIA, Switzerland