SAI VIGNESH

Building AI Systems in Public

LangChain • LangGraph • RAG • Python • Java

Learning Every Day 🚀

[SAI VIGNESH @SAIVIGN34633771]

Day 47
Added RAG Caching with Spring AI.
Cache HIT/MISS,Question normalization,Skip repeated RAG + LLM calls,Faster responses,Build smart, not just bigger.
#Java #SpringAI #RAG #GenAI #AIEngineering

[SAI VIGNESH @SAIVIGN34633771]

Day 46 -Upgraded my Day 44 RAG Observability project with resilience & fallback.
Jina Reranking failure handling, PGVector fallback
Production AI isn't just about accuracy — it's also about handling failures gracefully.
#Java #SpringAI #RAG #GenAI #AIEngineering

[SAI VIGNESH @SAIVIGN34633771]

Day 45
Upgraded my RAG system with Jina AI Reranking + Spring AI.
PGVector → Top 10 candidates
Jina Reranker → Top 3 relevant docs
Qwen → Grounded answer
Continued Day 44 observability & metrics
Retrieve broadly. Rerank precisely. Generate reliably.
#Java #AIEngineering

[SAI VIGNESH @SAIVIGN34633771]

Day 44 of #100DaysOfAIEngineering

Learned & implemented RAG Observability with Spring AI.

Tracked retrieval, LLM & request latency using Micrometer + Actuator.

Building RAG is one thing. Monitoring it in production is another.

#Java #SpringAI #RAG #GenAI #AIEngineering

[SAI VIGNESH @SAIVIGN34633771]

Day 42–43 of #100DaysOfAIEngineering

Learned & implemented RAG Evaluation with Spring AI.
Precision: 0.67 | Recall: 1.0 | MRR: 1.0
Big takeaway: Building RAG is half the job. Measuring it makes it production-ready.
#Java #SpringAI #RAG #GenAI

[SAI VIGNESH @SAIVIGN34633771]

@tanujDE3180 I will go with the Design B which scales better from Design A as Design A generates multiple End Points

[SAI VIGNESH @SAIVIGN34633771]

Day 41 — Hybrid RAG with Spring AI
Built a Hybrid RAG pipeline combining PGVector semantic search + PostgreSQL keyword search, then passing the combined context to Qwen for grounded answers.
41 days closer to becoming a Java AI Engineer.
#Java #SpringAI #RAG #GenAI #AIEngineering

[SAI VIGNESH @SAIVIGN34633771]

@shivangitech_97 Hi lets connect
https://t.co/ikFm3f7SXU

[SAI VIGNESH @SAIVIGN34633771]

Day 40 — Metadata Filtering with Spring AI
Built a RAG pipeline where the LLM automatically predicts the category and uses it to filter relevant documents from PGVector.
Also learned how similarity threshold affects retrieval quality.

#Java #SpringAI #RAG #GenAI #AIEngineering

[SAI VIGNESH @SAIVIGN34633771]

Day 38 — Conversational RAG completed! Combined ChatMemory + Neon PGVector + Spring AI to build a conversational RAG system with persistent document retrieval and context-aware responses. One step closer to becoming a Java AI Engineer. 🔥 #Java #SpringAI #RAG #GenAI

[SAI VIGNESH @SAIVIGN34633771]

Day 37 - Persistent RAG completed!
Built persistent RAG with Spring AI + Neon PostgreSQL + PGVector, using Ollama embeddings and Groq/Qwen for grounded answers with source attribution.
One more step toward production-ready Java AI engineering.
#SpringAI #Java #SpringBoot #RAG

[SAI VIGNESH @SAIVIGN34633771]

🚀 Day 36 — RAG with Real Documents completed! Built PDF-based RAG with document loading, chunking, Ollama embeddings, vector search, source attribution, and grounded LLM responses using Spring AI + Groq/Qwen. #SpringAI #SpringBoot #Java #GenAI #RAG #AIEngineering

[SAI VIGNESH @SAIVIGN34633771]

Day 35 — Spring AI RAG completed! Built a working RAG pipeline with Spring AI, Groq/Qwen and Ollama, covering embeddings, vector storage, similarity search, context retrieval, and context-based LLM responses.
#SpringAI #SpringBoot #Java #GenAI #RAG #AIEngineering

[SAI VIGNESH @SAIVIGN34633771]

Day 34 — Spring AI Chat Memory

Built a conversational AI that remembers previous messages using ChatMemory and conversation IDs.

Learning how AI assistants maintain context across interactions.

#SpringAI #Java #AIEngineering #BuildInPublic

[SAI VIGNESH @SAIVIGN34633771]

Day 33 -Spring AI: Structured Output
Today I learned how to make an LLM return structured data instead of plain text.
Built a TopicResponse Java record and used Spring AI's .entity() to map the AI response directly into a Java object.
#SpringAI #Java #AIEngineering #BuildInPublic

[SAI VIGNESH @SAIVIGN34633771]

https://t.co/kXdcLdRCgb

[SAI VIGNESH @SAIVIGN34633771]

Day 32 — Started with Spring AI
Today I moved from LangChain4j to Spring AI and built my first AI API using:
• Spring Boot + Java 17
• Spring AI ChatClient
• Groq
• System & dynamic prompts
• Temperature control
• REST API
Small step, but excited to build more with SpringAI

[SAI VIGNESH @SAIVIGN34633771]

Day 31

Started learning Spring AI today.

Covered ChatClient, models, prompts, system/user messages, and Spring AI architecture.

Coming from LangChain4j, the concepts felt familiar, but I learned how Spring AI fits into the Spring Boot ecosystem.

#SpringAI #AIEngineering

[SAI VIGNESH @SAIVIGN34633771]

Github link:https://t.co/Fl0an3NrVk

[SAI VIGNESH @SAIVIGN34633771]

Day 30
Today I finished my first working RAG pipeline with Spring Boot + LangChain4j.
I worked on document loading, chunking, embeddings, vector search, and finally connected the retrieved context with Groq to generate the answer.
Next up Spring AI
#AIEngineering #BuildInPublic