Ashish Pise

AI/ML | GenAI | RAG | MLOps | Building

[Ashish Pise @offbeat_ash_18]

Netflix Movie Recommendation Engine-finally done!

Notebook → full ML Engineering + MLOps project
SVD+Popularity ensemble | FastAPI | Docker |51 tests | CI/CD |MLflow

RMSE 0.9189 | P@10 0.1378

Biggest lesson: the model is only one part of the system.
https://t.co/wY37Gr69F3

[Ashish Pise @offbeat_ash_18]

@riteshs2907 Yes, consistency is the key

[Ashish Pise @offbeat_ash_18]

Just unlocked the 100 Days Badge 2026 🏆
• 90+ days of DSA + SQL
• 184 problems
• Rank: 971,892 (under 1M finally)
Major patterns covered in Python. SQL getting stronger every week.
Still grinding. 📈

#LeetCode #DSA #Consistency

[Ashish Pise @offbeat_ash_18]

@d0znpp Impressive loop. Getting 217 agent-generated reports is a much cleaner signal than pure human feedback for performance work.

[Ashish Pise @offbeat_ash_18]

My work in offbeatash/Netflix-movie-recommendation-engine was selected for the XERJ Corpus Hub.
AI coding agents can now retrieve it as a real worked reference instead of guessing.

https://t.co/XCcl9QJ4dF

#ML #OpenSource

[Ashish Pise @offbeat_ash_18]

🚀 XERJ reached out to feature my Netflix Recommendation Engine in its Corpus Hub, helping AI agents learn from real code!

🔗 https://t.co/SOH2wQlwEj

#AI #OpenSource

[Ashish Pise @offbeat_ash_18]

Check out xerj repo: https://t.co/qSozkYkIJV

[Ashish Pise @offbeat_ash_18]

🚀 XERJ reached out to feature my Netflix Recommendation Engine in its Corpus Hub, helping AI agents learn from real code!

🔗 https://t.co/SOH2wQlwEj

#AI #OpenSource

[Ashish Pise @offbeat_ash_18]

Connected GenAI components into a real workflow 🚀

Built a LangGraph RAG agent that picks its own tools:

Gemini + FAISS → RAG → Tavily + custom tools → LangGraph

User → LLM → Tool → Result → Answer

RAG = knowledge
Tools = power
LangGraph = control

#RAG #langgraph

[Ashish Pise @offbeat_ash_18]

🛠️ Tool Binding & Calling with LLMs
Focused on the full loop:
LLM → Tool Call → Execution → ToolMessage → LLM → Final Response

Bound multiple tools to a single LLM.

Real power isn’t just access to tools — it’s the interaction loop.
Step forward in Agentic AI ⚙️

[Ashish Pise @offbeat_ash_18]

@HosseinGorji021 More to learn and build🤝

[Ashish Pise @offbeat_ash_18]

🛠️ Tool Calling with LangChain
Built custom tools, async tools, StructuredTool + Wikipedia/Tavily.
Key shift:
LLM → Tool → External info → LLM

Building block for Agentic AI.
Next: Agents, RAG, LangGraph.

🔗GitHub: https://t.co/r2EWWTkNgI

[Ashish Pise @offbeat_ash_18]

Built a simple QnA chatbot using groq inference and streamlit.

Much more to build
GitHub: https://t.co/r2EWWTkNgI

#chatbot #GenAI

[Ashish Pise @offbeat_ash_18]

@jiyajain_23 had the same experience first time installing it

[Ashish Pise @offbeat_ash_18]

Building in public.
Breaking things. Fixing them. Learning GenAI. 🤖⚡
gen-ai-lab is just getting started.

GitHub: https://t.co/r2EWWTkNgI

#GenAI #AI #LangChain #LangGraph

[Ashish Pise @offbeat_ash_18]

@AnshikaK7 https://t.co/ktljz14x09

[Ashish Pise @offbeat_ash_18]

Tried uv for Python project management and I'm impressed by how much it simplifies the workflow.

Environment creation, dependency management, locking and running tools feel much cleaner than my usual pip + venv workflow.

Another tool going into the toolbox.

[Ashish Pise @offbeat_ash_18]

🚀 My Netflix Recommendation Engine is now live on @HuggingFace Spaces!

Live demo: https://t.co/9p2wc9haVH

GitHub: https://t.co/SOH2wQlwEj

#MachineLearning #MLOps #RecommenderSystems #HuggingFace #Gradio

[Ashish Pise @offbeat_ash_18]

@imthepk https://t.co/Sl3Hxjb6b9

[Ashish Pise @offbeat_ash_18]

Check out my clean AI/ML portfolio🔥

Portfolio → https://t.co/imqWAOnrj3

#AI #MachineLearning #MLOps #BuildInPublic

[Ashish Pise @offbeat_ash_18]

Just crushed SQL 50 on @LeetCode 🔥
Feeling that database glow ✨
#LeetCode #SQL50 #SQL #CodingJourney