Skip to content
View Asif-0209's full-sized avatar

Block or report Asif-0209

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.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. 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
Asif-0209/README.md

Asif Pinjari

AI engineer and educator who builds, teaches, scopes, and translates. I ship production AI solutions with Claude and turn non-technical colleagues into confident AI users.

I have built end-to-end agents, automations, internal tools, evaluation harnesses, and integrations on Claude API, LangChain, and AWS serving 200+ faculty and administrators across a public university, and run trainings for 180+ people, from middle-school summer campers to graduate students to senior faculty. I am energized by ambiguity, resource constraints, and wearing multiple hats, and I exercise judgment to catch wrong answers and push back when something should not be built.

Anthropic AI Fluency and Claude 101 certified · AWS dual-certified · IEEE-published researcher · PyPI open-source author · M.S. Computer Science, NAU (3.9 GPA)

Portfolio · LinkedIn · asifp0203@gmail.com


What I work on

Building with Claude. Agents, automations, and internal tools on the Claude API and Anthropic SDK: MCP, tool use, structured outputs, and agentic workflows that non-technical people can actually operate after handoff.

Agentic systems and evaluation. Multi-agent orchestration with LangChain and LangGraph, plus the evaluation harnesses (LLM-as-Judge, RAGAS, LangSmith) that turn "did the AI work?" into concrete metrics stakeholders trust.

RAG that stays grounded. Retrieval pipelines with inline citations and judgment layers that refuse low-confidence answers and flag policy-sensitive questions for human review.

AI enablement. Runbooks, onboarding guides, and training sessions that turn colleagues into the AI go-to people on their teams.


Experience

AI/ML Engineer — Northern Arizona University (Jan 2025 – Present) Built and deployed agents, automations, internal tools, and integrations on Claude API + LangChain + AWS, including RAG pipelines, LLM-powered dashboards, and agentic workflows used by 200+ faculty and administrators. Improved model accuracy ~30% and reduced token usage ~40% through prompt engineering, retrieval tuning, and selective model routing.

Research Assistant, AI Agents & Deep Learning — NAU (Feb 2026 – Present) Applied research on observability and anomaly detection in multi-agent AI communication systems; preparing peer-reviewed publication on agentic AI observability and self-refinement patterns.

Teaching — NAU (Jan 2025 – Present) TA for Python & Deep Learning (100+ grad students) and Advanced Python (80+ grad students); Semester Zero peer mentor teaching responsible AI use to middle-school students.


Projects

EthicAI — Classroom AI Literacy Toolkit for K-12 Educators Prompt-coaching assistant, AI-output verifier, and teacher dashboard with guardrails via system prompts, Pydantic structured outputs, and content filtering. 96% appropriate-refusal rate across 200+ test prompts. Claude API, LangChain, Streamlit.

OfficeHours — Internal AI Assistant for University Workflows RAG pipeline (Amazon Bedrock embeddings + FAISS) grounding every response in verified institutional documents with inline citations, plus a judgment layer that refuses low-confidence answers. 92% accuracy on a 150-question test set; serves 200+ users. Claude API, Bedrock, LangChain.

smart-preprocess · pip install smart-preprocess-asif Open-source PyPI library automating ML data preprocessing. Built and maintained solo with versioned, CI-driven releases.

Human Activity Recognition (CNN+LSTM) · IEEE InC4 2024 92% accuracy classifying 15 human activities from 1.2M+ sensor readings; published conference paper.


Stack

Build with Claude

  • Claude API, Anthropic SDK, Claude Code
  • MCP (Model Context Protocol), Tool Use, Structured Outputs, Agentic Workflows

Agentic AI & LLMs

  • LangChain, LangGraph, Multi-Agent Systems, Agent Orchestration, ReAct
  • RAG Pipelines, FAISS, Pinecone, Hugging Face, Amazon Bedrock
  • LLM Evaluation, LLM-as-Judge, RAGAS, LangSmith, Agent Observability

ML / Deep Learning

  • PyTorch, TensorFlow, Scikit-learn, XGBoost
  • CNNs, LSTMs, Transformers, NLP, Fine-Tuning (LoRA)

Cloud & Deployment

  • AWS (Lambda, S3, EC2, API Gateway, Bedrock, SageMaker, EKS), GCP
  • Docker, Kubernetes, Serverless, CI/CD, GitHub Actions

Languages & Data

  • Python (Advanced, OOP), SQL, JavaScript, Bash
  • Pandas, NumPy, PySpark, PostgreSQL, MongoDB, FastAPI, Pydantic

Pinned Loading

  1. scikit-learn-contrib/imbalanced-learn scikit-learn-contrib/imbalanced-learn Public

    A Python Package to Tackle the Curse of Imbalanced Datasets in Machine Learning

    Python 7.1k 1.4k

  2. Teaching-Tools-Development/class_ticket_system Teaching-Tools-Development/class_ticket_system Public

    Python

  3. Teaching-Tools-Development/visual_API_workflow Teaching-Tools-Development/visual_API_workflow Public

    JavaScript

  4. university-ai-ml-engineer-projects university-ai-ml-engineer-projects Public

    Selected projects built at NAU during my M.S. and as AI/ML engineer in NAU

  5. asifflix asifflix Public

    JavaScript

  6. smart-preprocess-asif smart-preprocess-asif Public

    A lightweight, one-line data preprocessing toolkit for machine learning workflows.

    Python