Divyakush Punjabi

404 bio not found

Joined on Aug 13, 2026 https://www.divyakush.com

Divyakush Punjabi

Aug 15

Multimodal AI: how a text model learns to see

#ai #machinelearning #computervision #genai

3 min read

Want to connect with Divyakush Punjabi?

Create an account to connect with Divyakush Punjabi. You can also sign in below to proceed if you already have an account.

Create Account

Already have an account? Sign in

Divyakush Punjabi

Aug 15

Guardrails: keeping LLM systems from going off the rails

#ai #llm #security #genai

3 min read

Divyakush Punjabi

Aug 15

Evaluating LLMs: why 'it looks good' isn't a metric

#ai #llm #mlops #machinelearning

3 min read

Divyakush Punjabi

Aug 15

Chain-of-thought: making a model think before it answers

#ai #llm #reasoning #genai

3 min read

Divyakush Punjabi

Aug 15

Temperature, top-p, and sampling: controlling LLM randomness

#ai #llm #genai #machinelearning

3 min read

Divyakush Punjabi

Aug 15

Vector databases: how semantic search scales

#ai #database #search #machinelearning

3 min read

Divyakush Punjabi

Aug 15

The context window: an LLM's working memory and its limits

#ai #llm #genai #machinelearning

3 min read

Divyakush Punjabi

Aug 15

Tokenization: why an LLM can't count the R's in strawberry

#ai #llm #nlp #machinelearning

3 min read

Divyakush Punjabi

Aug 15

Prompt engineering that actually works (and what does not)

#ai #llm #prompts #genai

3 min read

Divyakush Punjabi

Aug 15

Fine-tuning vs RAG vs prompting: pick the right lever

#ai #llm #rag #machinelearning

3 min read

Divyakush Punjabi

Aug 15

Getting reliable, structured data out of an LLM

#ai #llm #api #backend

3 min read

Divyakush Punjabi

Aug 15

What 'agentic AI' actually means (past the buzzword)

#ai #llm #agents #genai

3 min read

Divyakush Punjabi

Aug 15

Self-attention, explained without the heavy math

#ai #machinelearning #transformers #deeplearning

3 min read

Divyakush Punjabi

Aug 15

Embeddings and semantic search, from the ground up

#ai #machinelearning #nlp #search

3 min read

Divyakush Punjabi

Aug 15

Why LLMs hallucinate — and the patterns that actually stop it

#ai #llm #genai #machinelearning

3 min read

Divyakush Punjabi

Aug 15

RAG, explained properly: how retrieval keeps an LLM honest

#ai #llm #rag #machinelearning

3 min read

Divyakush Punjabi

Aug 13

A warning one minute early beats a perfect post-mortem

#machinelearning #iot #edgecomputing #sensors

1 min read

Divyakush Punjabi

Aug 13

Multi-tenancy is a promise you keep in the database

#fastapi #postgres #backend #saas

1 min read

Divyakush Punjabi

Aug 13

A dumb protocol is a feature, not a compromise

#embedded #iot #arduino #python

1 min read

Divyakush Punjabi

Aug 13

Your ML accuracy might be quietly cheating

#machinelearning #datascience #mlops #python

1 min read

Divyakush Punjabi

Aug 13

Hardware doesn't get a second frame

#fpga #verilog #hardware #embedded

1 min read

Divyakush Punjabi

Aug 13

The commerce bug that charges you twice — and how to kill it

#payments #backend #api #reliability

1 min read

Divyakush Punjabi

Aug 13

Two payment gateways, one button the customer never questions

#react #payments #stripe #frontend

1 min read

Divyakush Punjabi

Aug 13

For a servo, a cloud round-trip is a lifetime

#edgeai #iot #computervision #embedded

1 min read

Divyakush Punjabi

Aug 13

You can't build intuition for something you can't see

#algorithms #react #webdev #learning

1 min read

Divyakush Punjabi

Aug 13

The stop-word that quietly breaks every voice smart-home

#python #nlp #iot #arduino

1 min read

Divyakush Punjabi

Aug 13

Recommenders throw away your single most useful signal: order

#machinelearning #pytorch #recsys #datascience

1 min read

Divyakush Punjabi

Aug 13

The 'do nothing' zone that made my camera track smoothly

#computervision #iot #embedded #robotics

1 min read

Divyakush Punjabi

Aug 13

Saturday MK1: an AI assistant that's a system, not a prompt

#ai #python #fastapi #llm

3 min read

Divyakush Punjabi

Aug 13

Making an LLM trustworthy over bank statements

#python #fastapi #llm #fintech

3 min read

Divyakush Punjabi

Aug 13

HomeSense: turning speech into physical actions, no app

#python #iot #nlp #arduino

4 min read

Divyakush Punjabi

Aug 13

AlgoVerse: watching algorithms think at 60fps in the browser

#react #typescript #algorithms #webdev

4 min read

Divyakush Punjabi

Aug 13

Guilded-Guild: recommending the next item with SASRec in PyTorch

#machinelearning #pytorch #python #react

3 min read

Divyakush Punjabi

Aug 13

Building Netra: an edge-AI camera that tracks you on its own

#computervision #iot #python #react

4 min read

DEV Community

Divyakush Punjabi

Badges

Writing Debut

Multimodal AI: how a text model learns to see

Want to connect with Divyakush Punjabi?

Guardrails: keeping LLM systems from going off the rails

Evaluating LLMs: why 'it looks good' isn't a metric

Chain-of-thought: making a model think before it answers

Temperature, top-p, and sampling: controlling LLM randomness

Vector databases: how semantic search scales

The context window: an LLM's working memory and its limits

Tokenization: why an LLM can't count the R's in strawberry

Prompt engineering that actually works (and what does not)

Fine-tuning vs RAG vs prompting: pick the right lever

Getting reliable, structured data out of an LLM

What 'agentic AI' actually means (past the buzzword)

Self-attention, explained without the heavy math

Embeddings and semantic search, from the ground up

Why LLMs hallucinate — and the patterns that actually stop it

RAG, explained properly: how retrieval keeps an LLM honest

A warning one minute early beats a perfect post-mortem

Multi-tenancy is a promise you keep in the database

A dumb protocol is a feature, not a compromise

Your ML accuracy might be quietly cheating

Hardware doesn't get a second frame

The commerce bug that charges you twice — and how to kill it

Two payment gateways, one button the customer never questions

For a servo, a cloud round-trip is a lifetime

You can't build intuition for something you can't see

The stop-word that quietly breaks every voice smart-home

Recommenders throw away your single most useful signal: order

The 'do nothing' zone that made my camera track smoothly

Saturday MK1: an AI assistant that's a system, not a prompt

Making an LLM trustworthy over bank statements

HomeSense: turning speech into physical actions, no app

AlgoVerse: watching algorithms think at 60fps in the browser

Guilded-Guild: recommending the next item with SASRec in PyTorch

Building Netra: an edge-AI camera that tracks you on its own