I advise or advised the following startups or companies. - Niantic Labs: They are the makers of Pokémon Go game. They make 3D maps and associated Visual Positioning System for localization.- Predibase: Fine-tune and serve any open-source LLM at low cost using Low Rank Adaptation (LoRA) technology.- Fore.ai: Evaluation suite for LLM applications.- Washington Post: The famed news corporation.- Tomorrow Health: Technology solutions for home-based healthcare for providers, suppliers, and patients.- Stellar Health: Recommendations designed to meet the new objectives of value-based healthcare delivery models.- Quarterhill: They cater to Tolling and Enforcement industries and handle over 2 billion toll transactions each year.At times, some of these roles have been close to full-time roles with deep engagement on engineering designs, engineering management, product and business strategy, roadmap planning, execution, org design, interviews, and leadership mentoring.

LMNT is an early-stage venture fund focused on robotics & automation, climate & decarbonization, and industrial tech. I source deals based on the fund's focus.I also perform due diligence on prospective portfolio companies. As part of this, I engage with the founders and their teams to understand the core innovation or idea, business model, product strategy, and engineering architecture. I research the market to validate the moat and risks, and create a thesis on whether to invest.

I led the Maps platform at Uber.Maps are existential for Uber and major parts of core trip/Eats flows are built on Maps experiences and systems: ETA (Estimated Time of Arrival), traffic, routes, pick-up and drop-off experience, destination Search experience and Uber-owned Places data, driver navigation, map tiles, map data, sensor fusion augmented live location, etc. Highlights of my contributions:- Maps had a positive impact on Uber's core business metrics, such as Gross Bookings (GB), trip cancellations, completed trips as well as experience metrics, such as ETA accuracy, reroute volume, trips with critical safety and nav data fixes, etc.- For ETAs and Map Search, two of the most important aspects, we were X% more accurate, YYx faster, and Zx cheaper than the best commercial consumer maps APIs. - The core ETA service was the highest load app service at Uber: X million QPS, YY trillion queries per year. Served at ZZ millisecond latency. Also launched a transformer-based deep learning model for ETAs with internal partners.- Launched an ML-based automated real-world change detection and map deployment model reducing the latency to 4 min from a vendor-dependent manual process taking ~6 months.- Created internal map data authoring and ML labeling platform for human operators that eventually led to the creation of Uber's Scaled Solutions service.I joined Uber in 2015 when the Maps org was very small. I was part of a small group of leaders who envisioned a multi-year strategy and executed on it. I ended up leading the entire engineering org. The org size was ~200.

I worked in the Google Maps team and managed three different areas during my time at the company. In reverse chronological order:I was responsible for Google Maps'​ User-Generated Content and Google My Business products. These were full-stack teams, including a machine learning team, middleware team, and mobile and web client teams. Responsible for redesigning these products to increase user engagement and retention, get higher-quality contributions, fight spam, and increase the coverage and quality of automatic and user-based moderation of contributions.For first three years, managed the Ground Truth Tools team that built the world's best map-making software to create Google Maps data from scratch. Founded and/or led massive-scale, complex, multi-year projects, such as lane mapping, 3D buildings, Indoor Maps, automated framework for detecting and fixing map errors, etc. The innovation in this software under my watch led to new Google Maps features, new country launches, and major increase in mapping speed.The org size was ~75.

deCarta was a startup and we made the mapping platform behind many successful mapping services, including Google Maps and Yahoo Maps. This platform was very well-known within the industry but relatively unknown outside of it.Led the evolution of sample source code into production-grade software products. Launched new product lines for data and associated services, thereby enabling the company to enter more markets and countries. Increased the number of products released per year by about 10 times. This led to better positioning for the company and increased customer satisfaction. Launched an automated distributed data build system to handle the additional load. Started as the first engineer of the data team. Grew the team over time to meet the business needs described above. Managed the server and content teams in the US and two outsourcing teams—one in China and the other in India.

Responsible for design and implementation of database indexes for Magellan's in-vehicle navigation product, 750NAV™. The database application ran on an embedded platform with ARM processor and μC/OS operating system. Developed tools on Windows platform to search the indexes. Led the specification and design effort for a spatial database API library. An incarnation of this device still runs in Hertz cars as the NeverLost product.

Information Engineer: member of a team that designed and engineered a data warehousing software integrating mainframe with a three-tier architecture utilizing an Oracle database and a web front-end. Involved in the design, development, and enhancements of Lotus Notes workflow applications and Microsoft Access database systems.Business Associate: an Account Representative for over 600 customers in General Motors Truck Group headquarters. Involved in customer communications and projects, such as site-moves and equipment upgrades.