My work

Entrepreneur, ML and Software Engineer.

Pomelo

Co-Founder, Feb 2026 - Present

Co-Founder, May 2024 - Jun 2025

Built an AI mock F1 visa interviewer using voice AI for prospective international students coming to the US.
Bootstrapped and profitable: 7,000+ users and 7,000+ mock interviews, with students practicing 3–5 times a day.
Real-time voice pipeline: Whisper (STT), Mistral-7B + Haystack prompted as a consular officer, and a custom TTS layer with Vosk for ASR.
Payments were broken for people in India and students would try paying with multiple cards (from their parents) to pay for our tool. That's when we knew we had product market fit.
Set up networking and databases. Trained neural networks and ran inference tests.
Designed and built our Next.js app and landing page.

ML Engineer, Oct 2024 - Present

Driving data federation across platforms for a single-truth KPI reporting service.
Worked on a GraphRAG agent to enable analysts to query market and competitive intelligence questions and get source-cited answers from frequently ingested, approved documents and sources. Used LlamaIndex and AWS GraphRAG toolkit.
Worked on Computer Vision ML infrastructure using AWS SageMaker, Docker, Databricks, and MLflow.
Improved product detection algorithm accuracy by 10% through systematic MLflow experimentation.

Software Engineer (Photogrammetry and Volumetrics), Jun - Jul 2025

Automated manual photogrammetry and volumetric visualization processes of 150+ Roman/Punic excavation trenches in Sardinia, Italy.
Built an end-to-end automated workflow from photographs from trenches to 3D models generated using CloudCompare, Blender, and Metashape.
Increased turnaround time by 70%+.
Enabled parallelization of processes.

Co-Founder, Jan 2023 - May 2024

Built call agents for customer service by hacking up a Go server to orchestrate MediaStreams between Whisper (STT), Vosk (ASR), TransformerTTS (TTS), Haystack and a basic Language Model.
Integrated GPT and tuned Whisper.cpp, cutting response time by 25%.
Built a Shopify plugin for the voice agent.

Co-Founder, Feb 2022 - Jan 2023

Designed and trained a TTS model. Used Whisper for STT. Helped design and train the Deepfake GAN model for AI avatars. Created an early MVP to demonstrate the idea to customers.
Built a custom liquid-cooled GPU server to train our TTS and voice-to-face models on our own compute.
Pivoted to voice-only agents based on user feedback.

Lead Software Engineer, Sep 2021 - Oct 2022

At PALP, I managed a technical team to develop NLP for data extraction to support a $250M Getty-funded Pompeii Modeling project.
Built a Data Validation pipeline to verify 100K+ complex data records with varying structure.
Shipped a Gatsby + React + Node App to encapsulate the complex project in a UX friendly product.
Worked collaboratively with stakeholders to resolve technical roadblocks.

Machine Learning Engineer, Jun 2021 - Aug 2021

At Nike, I designed and Implemented an ML pipeline to automate the creation of 50K+ product names/year on Nike.com
Developed a Seq2Seq Encoder-Decoder Model, with an Attention head using BERT + Tensorflow, Keras libraries.
Presented findings and recommendations to stakeholders through detailed reports and visualizations, driving data-driven decision making.

Software Engineer (FUll Stack), Jun 2020 - Aug 2020

At Fidelity Investments, I accelerated Fidelity's transition to a new Cloud platform by resolving performance bottlenecks through advanced CICD pipelines; improved deployment efficiency by 40% and reduced downtime by 25%.
Launched team efforts to implement SSO in the migrated Cloud platform for 3 core web apps via Agile.
Led every stage of a Software Development Life Cycle for web features including planning, implementing, testing, reviewing, deploying and maintaining.
Participated in agile development methodologies, code reviews, source control management, build processes, testing, and actively engaging in daily stand-up meetings and sprint planning sessions.

Co-founder and Software Engineer, Sep 2019 - May 2021

Built a prototype for a VR headset with a Jetson Nano, lenses, Unreal Engine and TensorRT.
Developed Computer Vision neural nets for embedded systems (TFLite/LiteRT) to use hands to control the interface.
Implemented a Whiteboard app with the ability to draw + Educational content.

Teaching Assistant, Jan 2020 - May 2020

Led discussion sections for 120+ CS sophomores and juniors the fundamentals of good programming methodologies including Memoization and Dynamic Programming.

NLP Research Assistant, Feb 2019 - Sep 2021

Developed NLP algorithms to extract from unstructured data to contribute to a database for the digital mapping of Pompeii for a $250M Getty-funded project, the Pompeii Artistic Landscape Project.