Suggestions
Anirudh Koul
Author of O'Reilly's Practical Deep Learning Book - AI@Pinterest & ML Lead @NASA FDL - UN/TEDx Speaker - Seeing AI founder
Anirudh Koul, pronounced as 'Cool,' is a seasoned professional with over twelve years of experience in Applied Research and Data Science, specializing in Artificial Intelligence, Deep Learning, Computer Vision, and Big Data Analysis. He is renowned as the founder of Seeing AI, a widely used assistive technology for the visually impaired. Anirudh has a track record of delivering features to over a billion users and has been a key figure in developing computer vision and deep learning solutions for Augmented Reality, Robotics, Speech, and Accessibility tools. With a remarkable presence in the tech community, he has won numerous awards, including recognition from prestigious organizations such as IEEE, CES, FCC, Cannes Lions, TIME magazine, and more, and has been praised by global leaders like Justin Trudeau and Theresa May. Anirudh is a NASA FDL AI Mentor and serves as a judge for CES Innovation Awards and Edison Awards. He has delivered talks at prominent events like TEDxSeattle, UN, NASA, Google, and major conferences on Artificial Intelligence. Anirudh's recent projects, including the Seeing AI app and hands-free typing in Microsoft Office, reflect his commitment to leveraging technology for social good.
Highlights
95 days after a fall, I hiked the elevation gain of Everest in 36 hours. I went from unable to go 1 floor pain free to 3000 floors in one go. Biggest factors that helped were luck, my coach Mark Postle and my physiotherapist Charlie Weng, DPT. This journey taught me that with the right support, mindset and some data science, anything is possible. Geek alert - Sharing behind the scenes look in case someone else wants to tackle a similar challenge:
Training:
- Spent daily hour of physiotherapy exercises after 2 knees, one ankle and one toe went out of order
- Progressed gradually to walking, spinning, elliptical, treadmill, hiking, hiking with weights.
- To train for altitude, went to Colorado, either hiking up Pike’s Peak (14K altitude) with 35 lb backpack (7-8 hours) or working remotely from the top.
- Did 3500 jumping jacks from the 14k feet (60% oxygen), as well as restricted breathing exercises to strengthen lungs
- For heat training, spent time in a Sauna getting comfortable to 196 F and eventually doing jumping jacks there too
Race Day Planning:
- Built an ML model (on data from previous hikes) to calculate race day performance and gradual slowness from fatigue
- Built a spreadsheet with different timing variables (e.g. seconds at each water station, washroom, time to hike up) to simulate different scenarios
- Calculated the minimum steps per minute needed. Built playlists at different Beats Per Minute, so depending on heat, fatigue, I could vary the songs and hit my hiking poles to the alternate beats of the music.
- Abstained from listening to these songs, so they give extra mental boost at the event
- Tried building a motorized fan with water spritzers to cool neck. Found neck fans is already a thing, so purchased from Amazon.
Race Day:
- Goal: Get to half Everest at 29029
- Logistics: Hike up, take gondola down, repeat 13 times
- Strategy: Don't stop, keep putting one step after the other, even if slowly.
- Last person to start, so no one could overtake me. But I could overtake many. This was a huge confidence boost.
- Surprise: Performed better than I anticipated. I reached my goal by the 9th hour. (6 laps)
- With no more goal pressure, I took a victory lap, then another, then another including during a thunderstorm … (8 times)
- Body pretty much moved in autopilot mode around 137 bpm that it was trained on for hundreds of hours
- At one point, I reached top 6 out of 339 athletes
- Downfall: Ate human food first time after 18 hours (instead of gels), stomach bloated, then couldn’t drink or eat much for next 7 hours. Glycogen in legs fully depleted. Required pausing, 2 breakfasts and 1 lunch to get back.
- Finished the Everest height
- Found a cowbell and did another lap to cheer those struggling up. Eventually finishing with a full 10000 meters, ~33000 feet elevation gain
"In times of crisis, we do not rise to the level of our expectations, but fall to the level of our training." This was exactly my experience
#29029everesting
