Skip to content
View Dhruv-Sharma29's full-sized avatar

Highlights

  • Pro

Block or report Dhruv-Sharma29

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Dhruv-Sharma29/README.md

~/ whoami

$ cat about.txt

Hi, I'm Dhruv Sharma, an AI Engineer and Data Scientist in training. I build production-grade machine-learning systems, from training and experimentation to deployment and monitoring.

Between Data and Decisions

  • Currently exploring machine learning, computer vision, LLM agents, MLOps, and data engineering
  • Portfolio: dhruvsharma29.vercel.app
  • Education: B.Tech CSE (AI & DS), Graphic Era Deemed University

~/ toolbox

tech stack

~/ contribution calendar

Dhruv Sharma contribution calendar



snake eating the contribution graph

~/ skill radar

skill radar chart language radar chart

~/ selected work

llm-inference-lab reliable-operations-agent
small-model-tool-use
project result stack
llm-inference-lab 1,800 requests, 0 errors; throughput up 14x to 16 concurrent Python FastAPI vLLM Prometheus
reliable-operations-agent Local 7B agent: 15/30 incidents vs 30/30 for a scripted runbook Python FastAPI PostgreSQL
small-model-tool-use Tuned 7B: 84–94% on unseen tasks vs 73% for a prompted 14B Python LoRA Qwen2.5 GGUF

旅は続いていく

Pinned Loading

  1. Ivy-1.5B Ivy-1.5B Public

    Small language model trained from scratch to turn device-control requests into exact tool calls. Includes a frozen tokenizer, a data pipeline, pretraining, tool-use fine-tuning, fail-closed evaluat…

    Jupyter Notebook

  2. llm-inference-lab llm-inference-lab Public

    OpenAI-compatible gateway for a locally served LLM that keeps latency predictable under load: bounded queue, explicit 429/503 rejections, deadlines, and generation aborted when the client disconnec…

    Python

  3. permission-aware-enterprise-search permission-aware-enterprise-search Public

    Search over company documents (keyword, vector and hybrid retrieval with a reranker, plus cited LLM answers) where every passage, citation, preview and cached result respects the user's permissions…

    Python

  4. reliable-operations-agent reliable-operations-agent Public

    Incident-response agent for a small jobs service: diagnoses alerts from evidence it actually observed, takes only policy-permitted actions with approval and an append-only audit trail, resumes afte…

    Python

  5. small-model-tool-use small-model-tool-use Public

    Can a LoRA-tuned 7B beat a prompted 14B at tool use? Scored by running its tool calls in a sandbox and checking the resulting state. Tuned 7B: 84–94% on unseen tasks over 3 seeds vs 73% for the 14B…

    Python

  6. Squint Squint Public

    Screen OCR for code on macOS: indentation, blank lines and brackets intact. On-device, free and open source.

    Swift 2