Apr 2022 – May 2024 · 2 yrs 1 mo

• Backed by a16z crypto, First Round Capital and Gokul Rajaram in a $6m seed round, Stelo Labs built tools to reduce fraud, phishing and scams for crypto wallet users.
• We built the Stelo browser extension, Stelo Explore (a way to create and share crypto transactions securely) and the Stelo API for wallets.
• Pivoted after consumer crypto adoption slowed to build Superpipe - an open source python framework for building, evaluating and optimizing LLM pipelines, and Superpipe Studio, an accompanying web UI, which was used by several VC-backed startups.

Mar 2021 – Feb 2022 · 11 mos

• Reading, learning, exploring, hacking, following my curiosity and writing about it at www.amandhesi.net

Jan 2020 – Present · 6 yrs 5 mos

• Scale AI, Databricks, Anthropic, Notion, Reducto, Farcaster, OnDeck, Highlight.xyz, Meow.co, Sleek (acq SnackPass), Composer, Bandit ML (acq)
• A mix of personal, syndicate, and secondary investments

Nov 2019 – Mar 2021 · 1 yr 4 mos

• DoorDash acquired Caviar from Square in November 2019.
• Built DoorDash's ads/promotions products and team. Did a little bit of everything: ML-based ranking and forecasting models, email automation systems, product strategy.
• Built DoorDash's marketing automation platform from scratch. The system automatically manages/optimizes DoorDash's $XXX M marketing budget. See blog post for more details.

Sep 2017 – Nov 2019 · 2 yrs 2 mos · San Francisco Bay Area

• Led all things Machine Learning at Caviar, a food delivery service owned by Square. Built and improved core components of Caviar's logistics systems - dispatch algorithm, ETA predictions, meal prep time predictions, fraud prevention. Also built our ML serving infra. Reduced average delivery times by several minutes.

Aug 2015 – Oct 2016 · 1 yr 2 mos

• Helped make Messenger the fastest growing app of 2015 through a combination of acquisition and retention efforts.

Jan 2015 – Oct 2016 · 1 yr 9 mos

• Started the Facebook Blood Donations product during a company hackathon. Along with an awesome team, designed, built, and conducted user research in our spare time. Today, it is a flagship product of Facebook's social good team.

Oct 2014 – Oct 2016 · 2 yrs

• Conceptualized, designed and launched the prototype for a user safety feature to prevent malicious profile impersonation.
• Built internal tools to crowd-source language translation data to improve Facebook's translation technology that serves xxxM MAU.

Jun 2013 – Aug 2013 · 2 mos

Jun 2013 – Aug 2013 · 2 mos

Feb 2013 – Apr 2014 · 1 yr 2 mos · Princeton, NJ

• Crowd-sourced language translation startup founded on campus.

May 2010 – Aug 2010 · 3 mos

May 2009 – Aug 2009 · 3 mos · Redmond