home
writing
talks
Writing
11 Apr 2026
Single Responsibility Agents
Do one thing. Do it well. — ML Trenches
19 Feb 2026
Strange memory usage
Memory profiling forward/backward passes in PyTorch — ML Field Notes
22 Jan 2026
Lumos: Inside Dream11's Leap from 50+ ML Models to One Foundation Model
Building a foundation model to replace 50+ specialized ML models — Medium / Dream11 Tech Blog
20 Jan 2026
Post training with RL
How RL adjusts inference path distributions in post-training — ML Field Notes
13 Dec 2025
Thinks and think-nots
Why LLMs risk letting developers bypass critical thinking — ML Trenches
17 Nov 2025
LLMs: The gift that keeps on giving
Emergent capabilities and test-time scaling — ML Trenches
12 Jul 2025
The AI Engineer's path
Evolution from PhD-required to democratized AI development — ML Trenches
30 Oct 2024
Uncertainty and all his friends
Aleatoric vs epistemic uncertainty in ML — ML Trenches
11 Jun 2024
Foundation FOMO
13 Nov 2023
The changing lines
25 Jul 2023
Running into the void
08 Jun 2022
Automation in credit underwriting: Indian MSME perspective
Automation's impact on Indian MSME credit — Protium
25 Jun 2022
Data is Discrimination
28 Feb 2022
Get yourself fired: An Employee's guide to automation
23 Jan 2022
Quotable Quotes - Margin Call
30 Oct 2021
On information, power and money lending
01 Jul 2021
Quotable Quotes - American Psycho
01 Feb 2020
The open source economy and social capital
Publications
PromptShap: Prompt Engineering with Game Theory
Nigam, 2026 — Under review
LUMOS: Large User MOdels — Scaling User Behavior Prediction with Foundation Models
Nigam et al., 2025
ForeCal: Random Forest-based Calibration for DNNs
Nigam, 2024 — CODS-COMAD 2024