Skip to content
Adhish Thite

Adhish Thite

CurrentlySr. AI Architect at Google

I help enterprises put Gemini and the rest of Google's AI stack into production. Most of my work starts where the demo ends: architecture, evaluation, latency and cost, the things that decide whether a model or a harness holds up in front of real customers. Before Google I led AI engineering at Elastic and spent four years building fraud, sign-up risk and identity-verification models at Synapse. I'm also one of OpenAI's Codex Ambassadors, testing pre-release models against real engineering work.

Email meRésumé
Based in
Pune, India
Experience
12+ years
Focus
Enterprise & Agentic AI
Portrait of Adhish Thite

Experience

Where I've worked

  • Google

    Jun 2026 – Present

    Mumbai, India

    • Working with enterprise teams to architect and deliver AI solutions on Google Cloud, centered on Gemini and Google's latest models.
    • Translating Google's AI research and platform capabilities into production systems that fit real business needs.

  • OpenAI

    Apr 2026 – Present

    Remote

    • Selected as an OpenAI Codex Ambassador, joining a small group of power users and community builders who support Codex adoption.
    • Test early and pre-release model and Codex changes, then share direct feedback with OpenAI's internal teams.
    • Evaluate new Codex releases against real engineering work and write about how I actually use them.

  • Elastic Co.

    Feb 2024 – May 2026

    San Francisco Bay Area [Remote]

    • Leading the design and deployment of innovative AI solutions that drive generative AI initiatives.
    • Steering teams to integrate advanced machine learning techniques with customer-centric software features.
    • Shaping product strategy and influencing the broader technical roadmap at Elastic.

  • Synapse Financial Technologies Inc.

    Nov 2019 – Feb 2024

    San Francisco Bay Area

    • Progressed from developing deep learning-based risk management and automated ID verification systems to leading full-stack, customer-facing ML features.
    • Spearheaded the design, deployment, and optimization of advanced pipelines for sign-up risk and fraud detection while mentoring teams and refining product strategies.
    • Collaborated with cross-functional stakeholders to ensure robust, scalable, and production-ready ML solutions across the organization.

  • Gemini Data Inc.

    May 2019 – Nov 2019

    San Francisco Bay Area

    • Engineered data science solutions to unlock enterprise value from complex data silos through sophisticated pipelines in Python and Apache Calcite.
    • Delivered actionable insights that boosted performance and enhanced real-time log analytics.

  • Accenture Services Pvt Ltd

    Apr 2015 – Jul 2017

    Pune, India

    • Contributed to the end-to-end delivery of eCommerce platforms through Salesforce customization and advanced Java development.
    • Led teams in automating testing processes, managing agile projects, and overhauling sales and service processes to drive operational excellence.
    • Delivered scalable, high-quality software solutions that enhanced both client and internal business processes.

Internships & research

  • Welch Labs

    Sep 2018 – Apr 2019

    Charlotte, NC

    • Designed ensemble models combining Convolutional SVMs and CNNs to eliminate false positives in windshield defect detection, collaborating with industry leaders like Microsoft.
    • Focused on reducing inference times and ensuring robust, production-ready ML solutions for quality assurance.

  • Zuora, Inc.

    Jun 2018 – Aug 2018

    San Francisco, CA

    • Developed ML models to streamline customer support by correlating support tickets with internal knowledge bases, leveraging both topic modeling and deep learning.
    • Enhanced access validation processes to accelerate ticket resolution and boost live support efficiency.

  • UNC Charlotte

    Feb 2018 – Jun 2018

    Charlotte, NC

    • Conducted advanced data analysis and visualization projects using Tableau and statistical methods to predict student outcomes and improve engagement.
    • Automated data extraction processes, significantly reducing analysis time and supporting operational decision-making.

  • Tata Research Development and Design Center

    Jul 2013 – Apr 2014

    Pune, India

    • Designed and implemented a web portal to manage internal parcel tracking, laying the foundation for comprehensive system design and lifecycle management.
    • Gained hands-on experience in full-stack development during a pivotal internship project, establishing a strong technical base for future roles.

Expertise

Architecture, agents, and production ML

Enterprise AI architecture and agentic systems today, built on a decade of machine learning in production.

  • Primary focus

    Enterprise AI Architecture

    Designing how AI fits into an enterprise: platforms, reference architectures, and the security, cost, and scale questions that decide whether a system ships.

    • AI solution architecture
    • Reference architectures
    • AI platform strategy
    • Security & identity
    • Cost & token economics
    • Scalability & reliability
    • Technical roadmaps
    • Customer architecture reviews
  • Primary focus

    Agentic AI

    Agents that plan, call tools, and remember, plus the harness, evaluation, and security around them.

    • Harness engineering
    • Harness optimization
    • Agentic evaluations
    • Multi-agent systems
    • Tool use & function calling
    • Model Context Protocol (MCP)
    • Agent memory & context
    • Agent observability
    • Agent security
    • LangGraph

  • Generative AI & LLMs

    Taking LLM applications from prototype to production on Gemini and other frontier models.

    • Gemini & Google Cloud AI
    • RAG & retrieval
    • Vector & hybrid search
    • Prompt engineering
    • Fine-tuning
    • LLM benchmarking
    • LLMOps
    • Prompt-injection defense

  • Machine Learning

    A decade of production ML before the LLM wave: risk, fraud, vision, and language models running in real products.

    • Fraud & risk modeling
    • Identity verification
    • Deep learning
    • NLP
    • Computer vision
    • Feature engineering
    • MLOps
    • Model deployment

Tools & platforms

Languages

  • Python
  • TypeScript
  • Java
  • Rust
  • SQL

AI frameworks

  • LangChain
  • LangGraph
  • LlamaIndex
  • PyTorch
  • TensorFlow
  • Hugging Face
  • scikit-learn

Cloud & data

  • Google Cloud
  • AWS
  • Azure
  • Elasticsearch
  • PostgreSQL
  • Redis
  • MongoDB
  • Convex

Engineering

  • FastAPI
  • Django
  • Next.js
  • React
  • Docker
  • Kubernetes
  • MLflow
  • CI/CD
  • Datadog

Selected work

Things I've built

Side projects I've shipped outside work.

MenuStat

2026

A tiny native menu-bar monitor for macOS. I wanted one that was free, light, and did exactly what I needed, so I built it with Codex.

  • Swift
  • SwiftUI
  • macOS
Website

Observatory

2026

A real-time dashboard for every token my OpenClaw agent burns, built after 457 million tokens turned into a $650 bill.

  • Convex
  • Next.js
  • Recharts
Live

GlucoWise

2025

A full-stack tracker for glucose, weight, meals, and workouts. A blood report scared me into building it when existing apps fell short.

  • Convex
  • Next.js
Website

Dyson MCP Server

2025

An MCP server that lets Claude control my Dyson air purifier over local MQTT, so "make the air quieter" actually does something.

  • Python
  • MCP
  • MQTT
GitHub

Education

Education & certifications

Degrees

Certifications

Show all 19

Contact

Have something in mind?

Email is the best way to reach me. A DM on X works too.

Email mejobs.adhish@gmail.com