Browse categories
English
$
USD
Connect to your AI
I will build a rag chatbot or ai chatbot trained on your documents with AWS bedrock
Devendra V
About this gig
I build a RAG chatbot that answers from your own documents, shows its sources, and says so when it does not know. I have 4.75 years running AWS in production.
Generic chatbots make things up. A RAG chatbot is an AI chatbot trained on your data through retrieval, so every answer stays grounded in your PDFs, docs and knowledge base.
What you get:
- Ingest PDFs, web pages, Notion, Google Drive or a database
- Vector or hybrid search with reranking
- OpenAI, Claude or Amazon Bedrock models
- API, web chat or Slack and Teams bot
- Deployed on AWS with auth and cost control
Recent result: a production RAG chatbot on AWS Fargate and Aurora where I raised retrieval accuracy from 26/31 to 31/31, at under 1 cent per question, with nightly evals.
Why me:
- Top Rated on Upwork with 130+ jobs completed
- HashiCorp Terraform Associate (003) certified
- 4.75 years of hands-on AWS DevOps work
Message me with your documents and use case before ordering so I can confirm the right package.
Get to know Devendra V
Devendra V
AI and DevOps Engineer, RAG, LLM Apps, AWS
- FromIndia
- Member sinceAug 2026
- Avg. response time1 hour
Languages
Hindi, English, Gujarati
I build AI systems that run in production, not demos that break in week one. RAG pipelines over messy internal docs, LLM apps, and agents that do real work.
The difference is what sits underneath. 4.75 years on AWS, Terraform, Docker and CI/CD means I handle deployment, monitoring and cost control, not just the model call.
Recent work: a RAG backend serving live client queries, monitoring stacks for multi-tenant hosting, and a large AWS bill cut without downtime.
Tell me what you want to build and I will tell you straight whether it is a two day job or a two month one.
My Portfolio
FAQ
Which vector database do you use?
pgvector on Postgres, OpenSearch, Pinecone, Qdrant or Chroma. For most clients pgvector keeps it simple and cheap.
Will my data stay private?
Yes. It can run in your own AWS account, and with Bedrock your data stays inside AWS. Nothing is used to train public models.
What do you need from me to start?
Sample documents or access to where they live, a short list of real questions users ask, and an AWS account or API key for the model you prefer. I send a checklist after you order and never need your AWS root password.
How long does it take and how many revisions are included?
Each package lists its delivery time and number of revisions. Large document sets or custom integrations can take longer, so I confirm the timeline with you before you order.
Are AWS and model API costs included in the price?
No. AWS hosting and model API usage are billed to your own accounts. I design for low running cost, and my own production RAG chatbot answers at under 1 cent per question.
