AI Engineer · Delhi NCR

Building things
that listen,
reason, and ship.

I build Generative AI and Agentic AI products end to end. Currently architecting AIRA at OVE, a real time voice interview agent live in production for clients in the US and global EdTech.

Download Resume Cover Letter Get in touch
statusavailable exp2.8 yrs noticeserving
me products channel companies skills
01 · about

An engineer first. A builder always.

I'm Yash. I build AI products that ship to real users, not slide decks. For the last 2.8 years I've been working at the intersection of LLMs, agents, and real time systems.

Right now at OVE I'm building AIRA, a voice AI that conducts live interviews for US and global EdTech clients. Before that, at AGL-Hakuhodo, I built RAG pipelines, LangChain agents, and predictive models that pushed team efficiency up by 90%.

On the side I co-founded Engineer Being, a channel and community where we break down how production AI systems actually get built. Not tutorials. Not hype. Just what works.

I move comfortably across the full stack: data pipelines, model layer, APIs, deployment, and frontend. Mostly Python and FastAPI, increasingly Kubernetes and MLOps. I'm currently growing into DevOps and MLOps engineering as a parallel track.

02 · now

Currently shipping.

● Live in production · v1
AIRA AI Real Time Interview Agent
LiveKit Deepgram ElevenLabs Groq LLaMA 70B FastAPI React Azure GitLab CI/CD

AIRA conducts live voice interviews end to end. It handles resume and JD parsing, ATS matching, candidate shortlisting, the actual voice conversation, real time face proctoring via BlazeFace, and post interview scoring across five dimensions.

Deployed for City Home (Florida's top furniture retailer) and GUS (global schools group). Multi tenant. Multi vertical. Replacing manual HR screening at scale.

Also shipping a Multimodal Knowledge Graph pipeline. Scanned PDFs and newspapers in, structured Neo4j graphs out, 85 to 94% extraction accuracy.

03 · selected work

Things I've built and shipped.

/01
AIRA Voice AI Interview Agent
OVE · Nov 2025 → Present · Live
Production grade real time voice agent that runs the full hiring loop. WebRTC, sub 800ms latency, multi tenant personalization, 5 dimension AI scoring, candidate proctoring. Replacing manual HR screening for US and global clients.
LiveKit · Deepgram
ElevenLabs · Groq
FastAPI · React
/02
Multimodal Knowledge Graph Builder
OVE · Nov 2025 → Present
Pipeline that turns scanned PDFs, newspapers, and image archives into structured Neo4j graphs. OCR plus computer vision plus multimodal LLMs for entity extraction. Fuzzy deduplication, interactive PyVis viz, 85 to 94% accuracy.
PaddleOCR · OpenCV
Gemini 2.0 · NetworkX
Neo4j · PyVis
/03
Agentic HR Virtual Assistant
AGL-Hakuhodo · 2024 → 2025
Multi tenant HR chatbot with RAG, JWT auth, and JSON context management. Handled leave, attendance, and policy queries. Cut HR workload by 50%. 96% query resolution accuracy across multiple client orgs.
LangChain · Flask
RAG · JWT
OpenAI · GCP
/04
SEO Insight Generator
AGL-Hakuhodo · 2024 → 2025
Streamlit chatbot over 50+ client files using OpenAI embeddings and All-MiniLM-L6-v2. Generates statistical insights and charts via natural language. Replaced Excel and Power BI for ad hoc analytics. 90% efficiency lift.
Streamlit · Pandas AI
LIDA · FAISS
OpenAI Embeddings
/05
RAG Pipelines + LangChain Agents
AGL-Hakuhodo · 2023 → 2025
Built RAG over 50+ client files with OpenAI embeddings. LangChain and LangGraph agents for tool execution and adaptive retrieval. Llama 2 chatbot on GCP Cloud Run cut response time by 60%. Boosted SEO team efficiency 50%.
LangChain · LangGraph
Llama 2 · GCP
Cloud Run · Docker
/06
Predictive ML Models Churn & Regression
AGL-Hakuhodo · 2023 → 2024
XGBoost and scikit-learn models on customer datasets. Feature engineering, ensemble methods, cross validation. Hit 85 to 88% accuracy and precision. Deployed via Flask APIs for product integration.
XGBoost · Scikit-learn
Pandas · Flask
04 · channel & community

Engineer
Being.

A YouTube channel and community I co-founded for people learning to actually build production AI. We break down voice agents, RAG, MLOps, deep learning, and the messy middle of shipping real systems.

6.73K
subscribers
74
videos
2023
founded
watch on YouTube →
EB.
@engineerbeing · est. 2023
05 · experience

Where I've been.

Nov 2025 → Present
AI Engineer
Optimal Virtual Employee (OVE)
Architecting AIRA, a production voice AI interview agent for global clients. Also building a multimodal Knowledge Graph pipeline. Owning backend (FastAPI), frontend (React/Vite), proctoring (BlazeFace), and CI/CD (GitLab → Azure).
Noida, IN
Oct 2023 → Nov 2025
Data Science Analyst
AGL-Hakuhodo India
Shipped GenAI chatbots, RAG pipelines, LangChain and LangGraph agents, and XGBoost predictive models. Improved team efficiency by 90% and cut HR workload by 50% via an agentic HR assistant. GCP Cloud Run deployments.
Gurugram, IN
2023 → Present
Co-Founder
Engineer Being
Co-founded a YouTube channel and community focused on production AI engineering. Voice agents, RAG, MLOps, and the realities of shipping AI systems that work.
Remote
06 · stack

Tools I reach for.

Languages

  • Python
  • JavaScript
  • TypeScript
  • SQL
  • React
  • NodeJS
  • C/C++

Generative AI · Agents

  • LangChain
  • LangGraph
  • RAG
  • AI Agents
  • Agentic AI
  • LLMs
  • Prompt Engineering
  • Fine Tuning
  • Vector Embeddings
  • Knowledge Graphs

Voice AI · Real Time

  • LiveKit WebRTC
  • Deepgram STT
  • ElevenLabs TTS
  • Cartesia
  • WebSockets
  • SSE Streaming

ML · Deep Learning

  • PyTorch
  • TensorFlow
  • Scikit-learn
  • XGBoost
  • Hugging Face
  • Transformers
  • BERT
  • CNN · RNN · LSTM
  • OpenCV
  • YOLO
  • BlazeFace

Backend · Frameworks

  • FastAPI
  • Flask
  • Streamlit
  • REST APIs
  • JWT Auth
  • Multi-tenant Architecture

Data · Storage

  • FAISS
  • Neo4j
  • NetworkX
  • MongoDB
  • MySQL
  • PySpark
  • Hive
  • Pandas · NumPy
  • OCR (PaddleOCR · Tesseract)

LLMs · APIs

  • OpenAI
  • Groq
  • Gemini
  • LLaMA
  • Claude
  • Hugging Face Inference

Cloud · DevOps · MLOps

  • AWS (SageMaker · Lambda · EC2 · S3)
  • GCP (Vertex AI · Cloud Run · Dialogflow CX)
  • Azure (VM · App Service)
  • Docker
  • Kubernetes
  • GitLab CI/CD
  • n8n
  • MLOps

Let's build
something.

I'm currently open to AI Engineer, GenAI Engineer, ML Engineer, and MLOps roles. Immediate joiner, serving notice. Also up for collaborations and Engineer Being guest spots.

email yashraj25118110@gmail.com linkedin in/rajyash github yash-raj15 youtube Engineer Being