Dec 2023 – Mar 2024 · 3 mos · San Francisco Bay Area

• participating in WWF Science Fund AI projects. edge AI open-source projects of personal interests related to active listening via ambient sensors/microphones, using tonality, and intent, and biomarkers. fused modalities with CV segmentation, and I/O.
• github

Nov 2022 – Jan 2024 · 1 yr 2 mos · San Francisco Bay Area · On-site

• [Plug-and-play Gen AI modules with the founding team of Tensorflow] - Team of 7 ex Google Brain, NVIDIA, Cerebras and Stanford Computer scientists.
• built low-cost, easy to integrate hardware wearable modules that bring ML capabilities like natural language understanding, gesture recognition, presence detection, and voice interfaces to everyday devices, without depending on cloud.
• Led AI pipelining effort (whisper in->LLM->TTS) with Llama, Vicuna7b, and Finetuned task specific Small Language Models. Running locally offline on sub 100$ hardware. Prototyped several use-case scenarios for B2B customer pilots.
• Liaised with investors, board, design-partners, open-hardware community science and hf models collaborators
• Led collaboration with several top open source communities, and Harvard Medical School to explore AI powered hearable wearables for dementia, and passive summarization via wearables with 24x7 speech signals.

Jun 2016 – Dec 2023 · 7 yrs 6 mos · Greater Boston Area

• [Can we design systems that utilize CO2 and carbon and make scalable material streams from it]we are a small team of engineers and scientists (electrochemistry, mechatronics) trying to build a tabletop CO2 capture system. We open-sourced a part of the project: https://www.hackster.io/air-synth/air-synth-162320
• Our brand AIR-INK, makes materials from captured CO2 used by several top automotive, and consumer brands. Mastercard, DELL, Heineken, Diageo, used by 10,000s of printers/artists across the world.
• https://hypebeast.com/2022/10/johnnie-walker-air-ink-collaboration-unboxing
• Example of partnerships: https://www.forbes.com/sites/brookerobertsislam/2021/04/20/pangaias-game-changing-collaboration-draws-carbon-from-air-pollution-to-print-textiles/
• Demo: www.graviky.com

Jan 2013 – Apr 2015 · 2 yrs 3 mos

• Led on organizing Asia's biggest innovation cohort to bring design, technology, and art students together at 5th MIT Media Lab India Initiative Design Innovation Workshop 2015. The workshop bought 25 MIT instructors from Cambridge/MA and 350 students from around the country, and 15 startups doing interdisciplinary work.
• More than 100 demo-able prototypes were created, and demoed. india.media.mit.edu

Aug 2012 – Sep 2015 · 3 yrs 1 mo

• Researched and built augmented reality computational interfaces using transparent OLEDs, gesture recognition, applied to manufacturing and architectural tools.
• During his time at MIT, Anirudh developed several projects that were published at the top Human Computer Interaction conferences such as CHI, UIST, TEI as well as featured in popular media: The New York Times, CNN, MIT Technology Review, The Times, Fast Company, BBC, Discovery Channel Etc.
• Part of Anirudh's work there, went forward to commercialization with various partners.

Mar 2010 – Mar 2019 · 9 yrs · On-site

• Implemented an on-device gait analysis and haptic feedback system to aid the visually-challenged navigate and later gait analysis+ wearable muscle stimulation . Different iterations of the technology and product used for fall prevention and rehabilitation.
• I led concept and technical development at HP Labs (microcontroller, sensor fusion, industrial design) and spunoff as Ducere, for haptic navigation, and piloted it with the first 20 visually challenged users in Bangalore/Hyd.
• Built agile team at a cross of data science, gait analysis, human factors, electro-mechanical engineering.
• Lechal insole gait stack is licensed to several shoemakers and healthcare companies.

Mar 2010 – May 2010 · 2 mos

• Enhancing browsing experiences through Firefox using new media interface. Wrote gesture recognition algorithms from ground-up that make the Mozilla Firefox browsing experience work on a mouse-less interface i.e. a multitouch tabletop display, augmented reality interface.

Feb 2010 – Apr 2012 · 2 yrs 2 mos · Bangalore

• Developed an adaptive CAD modeling interface based on speech+pose estimation and stereoscopic displays. Prototyped software, hardware and form factor of the final piece which is now displayed at HP Labs innovation center, Zurich
• Multimodal Interfaces
• New input methodologies
• Gestural Interfaces
• User evaluation

Apr 2009 – Feb 2010 · 10 mos

• I am working on extending Google SketchUp to gesture recognition and stereoscopy for the NUIGroup TouchLib and CCV as a part of GSoC 2009