what's currently in production
|
🧠 LLMs Pretraining quirks, fine-tuning, evals, and the occasional prompt that breaks everything. |
🤖 Agentic AI Multi-agent orchestration, tool use, memory, and knowing when an agent should not act autonomously. |
💬 Conversational Agents Dialogue systems that mostly don't hallucinate your return policy. |
|
📖 NLP Tokenizers to retrieval, information extraction to hybrid re-ranking. |
👁️ Computer Vision Detection, segmentation, and teaching machines to actually look. |
⚙️ MLOps The unglamorous plumbing that keeps a model alive past the demo. |
conferences, meetups & workshops: this one's on tape
|
NOW PLAYING · hosted by Python India
Building retrieval-augmented chatbots that don't choke on the open web: an async scraping pipeline feeding a hybrid re-ranking layer, for answers that are actually grounded. ▸ more appearances get added here as they happen. |
the toolbelt, in credits order
STARRING: core languages & frameworks
CO-STARRING: LLMs & agentic AI
FEATURING: NLP & conversational
ALSO STARRING: computer vision
SUPPORTING CAST: MLOps & infra
AND INTRODUCING: web & data
BEHIND THE SCENES: systems & tools
a researcher's watchlist has a tell
| Film | Why it stays with me |
|---|---|
| The Prestige (2006) | Obsession, method, and the cost of chasing the perfect result: every researcher's secret biopic |
| The Shawshank Redemption (1994) | Patience... Some breakthroughs take nineteen years and a rock hammer. |
| The Truman Show (1998) | "We accept the reality we're presented." Until you start questioning your training data |
| 12 Angry Men (1957) | One dissenting opinion, followed to its logical end |
LinkedIn · X · Medium · Substack · Email
🎬 Behind the Scenes
The portrait above is real ASCII art, luminance-mapped from a photo through a hand-written Python/PIL script,
not a filter or a plugin. See scripts/ if you want to point it at a different photo.
