Akshat Agarwal

Co-Founder

San Francisco, California, United States9 yrs 10 mos experience

Highly Stable

Key Highlights

Expert in multilingual LLM development at Meta.
Led machine learning initiatives in autonomous driving.
Co-founded a successful educational robotics startup.

Stackforce AI infers this person is a Robotics and Machine Learning expert with a strong entrepreneurial background.

Contact

agarwalaks30@gmail.com LinkedIn

Skills

Core Skills

Machine LearningResearchComputer VisionRobotics

Other Skills

AlgorithmsArduinoAutodesk InventorCC++Deep LearningEmbedded Operating SystemsEmbedded SystemsImage ProcessingLaTeXLeadershipLinuxMatlabMicrosoft OfficeMobile Robotics

About

I’m working on improving the multilingual capabilities of Llama models and Meta AI, as a research scientist in the Meta Superintelligence Labs (MSL). Previously spent a year working on making Instagram Reels recommendations more timely and culturally relevant. Before that, I spent 5 years working in computer vision at Nuro, an L4 autonomous driving startup, working on object tracking and classification. I got my Masters in Robotics from CMU while doing research in multi-agent reinforcement learning with Prof. Katia Sycara. I completed my undergrad from the Indian Institute of Technology, Kanpur, during which time I co-founded an educational robotics hardware company. My website agakshat.com has more details

Experience

9 yrs 10 mos

Total Experience

2 yrs 6 mos

Average Tenure

2 yrs 2 mos

Current Experience

Nuro

4 roles

Staff Machine Learning Engineer

Oct 2023 – Mar 2024 · 5 mos

Back to building after couple years managing a team

PythonPyTorchC++Computer VisionMachine Learning

Senior Tech Lead Manager

Promoted

Dec 2022 – Oct 2023 · 10 mos

My team uses machine learning (computer vision and sequence modeling) to perform object tracking and fine-grained attribute classification. We also work on OOD (out of distribution) understanding, and trajectory prediction from raw sensor data.

machine learningcomputer visionsequence modelingobject trackingattribute classificationOOD understanding+3

Tech Lead Manager, Perception

Dec 2021 – Dec 2022 · 1 yr

Machine Learning Engineer

Jul 2019 – Dec 2021 · 2 yrs 5 mos

Carnegie mellon university

Graduate Research Assistant

Sep 2017 – May 2019 · 1 yr 8 mos · Greater Pittsburgh Area

My thesis was on robust and transferable multi-agent reinforcement learning using graph neural networks.
Previously worked on improving interpretability of deep reinforcement learning, developing safer AI and improving dialog quality in visually grounded dialog agents through community regularization.

Agilo technologies

Co-founder and CTO

Aug 2015 – Dec 2016 · 1 yr 4 mos · Kanpur Area, India

Spearheaded the development of evive, an open-source electronics platform for learning and prototyping electronics circuits, building robots and hacking together IoT solutions. It was successfully crowdfunded on Indiegogo in September 2016, raising $34k from 300+ backers from over 30 countries.
evive won the Hackaday Prize Automation Challenge, and has been featured in more than 25 media portals. For more information, see http://igg.me/at/evive

multi-agent reinforcement learninggraph neural networksdeep reinforcement learningdialog quality improvementRoboticsMachine Learning