20th July, 2026

What am I working on?

On a travel collage from my recent trip to Udaipur and Kumbhalgarh. Did you know Kumbhalgarh has the longest wall in India — second only to the Great Wall of China?

Here is the collage. I have made a couple of collages of all my recent travels - you can find them on my Instagram profile.

What am I reading?

I read They Called Us Enemy, a graphic memoir by George Takei (co-written with Justin Eisinger and Steven Scott, and illustrated by Harmony Becker) It is a deeply personal account of the incarceration of Japanese Americans following the attack on Pearl Harbor during WW2. Through the dual lens of his childhood memories of living in the camps (in some panels he is excited about the prospect of living in a camp as he is just a child who does not realise the gravity of the situation) and his adult understanding of the political situation, Takei talks about the trauma of being labeled an "enemy of the state" and forced from his home.

Takei talks the four years his family spent in internment camps and about the systemic indignities his family faced including mandatory loyalty questionnaires, pressure to renounce their citizenship despite being American-born, painful choices about volunteering for military service while their families remained in camps, and the tragic loss of known people in the Hiroshima attack.

Unlike graphic novels I have read that use experimental art styles, They Called Us Enemy uses clean, straightforward black & white panels to tell the story. This simple approach allows the gravity of the narrative to take center stage, creating a powerful and accessible historical document. Before reading this novel, I was not aware this is what Japanese Americans had to face during WW2. The U.S. formally apologized for Japanese internment camps through the Civil Liberties Act of 1988, which was signed into law by President Ronald Reagan.

What am I watching/watched recently?

Flee is a Oscar-nominated animated documentary that tells the harrowing true story of Amin Nawabi (a pseudonym), an Afghan refugee at the intersection of past trauma and his present life as a gay. Amin chronicles the arduous journey he undertook as a child, fleeing Afghanistan with his mother and siblings just before the country descended into total turmoil. Their escape leads them to a secluded, restrained life in Moscow, where they live in the shadows while desperate to secure passage to Europe through human traffickers.

The film’s visual language is new for me: it uses hand sketched, rough illustrations interspersed with real footage. The animation mirrors Amin’s psychological state; when the story turns to the most traumatic memories, the visuals shift into black & grey shadows, showing his desire to suppress the agonizing details of those encounters. As he talks about his childhood memories, the animation starts taking form as if we are witnessing these images being gradually coalesced within his mind in real-time.

Flee is a piece of animation that demands to be seen. You will feel the ache after watching this one.

New thing I learnt lately?

There are seven summits in the world which are the highest mountains on each of the seven continents. Puncak Jaya is the highest mountain in Oceania and one of the most topographically isolated in the world. They are tropical glaciers (yeah, you read it right), which are exceptionally rare. Most people associate the tropics with rainforests and heat, not ice, but the extreme altitude of Puncak Jaya (over 16,000 feet) allows ice to persist despite their equatorial location. However, they are disappearing rapidly (ofcourse, global warming) and projections suggest they will likely vanish within the next decade.

I also came across the bias-variance balance theory. Bias refers to errors introduced in the model due to overly simplistic assumptions. Variance measures how much the model’s predictions change if different training data were used. Put simply, Bias is like reading the heading and not the content inside while Variance is like memorising the text and not being able to answer anything beyond the text as you did not get the concept right.

To create models that are both accurate and reliable, one must check the bias-variance tradeoff. Underfitting occurs when a model has high bias due to overly simplistic assumptions and fails to capture the underlying patterns. You can fix this by increasing model complexity or adding relevant features. Overfitting occurs when a model has high variance and becomes overly sensitive, memorizing noise in the training data rather than the general trend. This is addressed by increasing the volume of training data or utilizing ensemble methods like Random Forest (that average multiple models to cancel out individual errors)