The OCR & Speech Workspace is built on two Mistral AI capabilities: OCR that turns a scanned image or a 400-page textbook into structured, searchable markdown, and real-time speech-to-text with a post-recording accuracy refinement pass.
Page-batched, not page-by-page
OCR runs concurrently across page batches rather than one page at a time, with live per-page progress streamed over SSE. A book-length scan doesn't mean a coffee break — it means a progress bar that actually moves.
Chat that cites the page it read
Once a document is OCR'd, it becomes queryable — document-scoped RAG chat with citations that point back to the specific page a claim came from. The pattern is the same one from RAG.NextUpgrad: an answer without a citation is a claim you can't check.
Live transcription, refined after the fact
Speech-to-text streams live as you talk, then a second pass over the completed recording corrects errors the real-time pass couldn't catch — the same tradeoff every voice product makes between latency and accuracy, made explicit instead of hidden.
Repo and architecture: github.com/Pranjulrathour/OCR-STT-NEXTUPGRAD-mistral.ai-.
GenAI engineer and AI product builder with 2+ years shipping production-grade AI systems: RAG pipelines, fine-tuned LLMs, hybrid retrieval and multi-modal apps across vision, speech and OCR, architected end to end from ingestion to deployment. Leads engineering for SCULT INDIA's 14-member team, founded the 500+ member TechVerse Enclave community and has mentored 200+ students. Three hackathon first prizes: Changethon 2025 (IIT Roorkee), Product Genesis at Vividhotsava 2025 (CSJMU Kanpur) and BYTEBATTLE (MeetKats).
Open to GenAI roles, hackathon judging, mentorship sessions and guest talks at colleges. Email pranjulrathour41@gmail.com.
