DEPLOYED FOR DELIVERAIL
EMMA
An AI voice receptionist that runs entirely on device. Speech in, grounded reasoning, speech out, with no per request cost.
faster-whisper · Ollama · Coqui TTS
SHEET 01
DWG. 001 / APPLIED AI ENGINEER
AI that
gets used.
I am Krishna Dadhich, an applied AI engineer with an MCA in AI and ML from CMR University. I build RAG systems, voice agents and conversational AI that real people use, from an offline voice receptionist deployed at a logistics company to a job search platform backed by 196 passing tests.
Today I work as an AI Automation Engineer at Deliverail, and I am looking for opportunities in GenAI, RAG, conversational AI or computer vision where I can keep shipping.
external API calls in EMMA
passing tests in CareerOS
live product sections shipped
DWG. 002 / SERVICE RECORD
Where it ran.
Dec 2025 to now
AI Automation Engineer
Deliverail, Bengaluru
LLM booking chatbot and an admin dashboard with SQL data management that reduced repetitive manual work.
Sep to Nov 2025
AI/ML Intern
Medha Tech Solutions, Bengaluru
MIRA, a domain restricted RAG chatbot with source grounded answers, plus role based admin in PHP and MySQL.
2024 to 2026
MCA, AI and ML
CMR University
CGPA 8. Certified by Anthropic, IBM and Google Cloud.
DWG. 003 / PROJECT SHEETS
Four systems, drawn to completion.
TESTED AT CMR UNIVERSITY
Kuro
A private multi agent assistant. Unlimited assistants, each with its own knowledge base and cited, streamed answers from your files.
React · FastAPI · Ollama · FAISS
SHEET 02
HUMAN APPROVAL FIRST
CareerOS
A personal command center that syncs Gmail, scores matches against my resume and tracks every application through a pipeline.
Next.js · TypeScript · Prisma · PostgreSQL
SHEET 03
RUNNING IN PRODUCTION
KENNY
An autonomous AI streamer persona that answers real comments and messages live on Instagram and Discord.
Python · Instagram API · Discord API · LLM
SHEET 04
DWG. 004 / KURO RETRIEVAL PIPELINE
From file to cited answer.
01
Ingest
PDF, DOCX, TXT, CSV, Excel
02
Index
FAISS vectors per assistant
03
Retrieve
Hybrid search with reranking
04
Generate
Local LLM through Ollama
05
Cite
Streamed answer with confidence score
DWG. 005 / MATERIALS
The toolkit.
PythonTypeScriptSQLPyTorchTensorFlowScikit-learnOpenCVLangChainLangGraphRAGFAISSOllamaFastAPINext.jsReactPrismaPostgreSQLMySQLDockerAzureMLflown8n