Purushotham Rao Eravalli profile photo 1 on Tal

Purushotham Rao Eravalli

AI/ML Research Scientist

Hyderabad

10+ years of work experience

Jukshio
Jun 20206 yrs 4 mosPresent

Hyderabad, Telangana, India

• Built production RAG pipelines with hybrid dense-sparse retrieval and LLM grounding for insurance document Q&A, clause extraction, and risk summarisation • Architected multi-agent retail operations platform using CrewAI, LangGraph, and AutoGen — For daily customer interactions, autonomous query resolution, inventory management, and escalation routing. • Fine-tuned LLMs (LLaMA, Qwen, DeepSeek, Mistral) from scratch and with LoRA/QLoRA/PEFT/DeepSpeed for domain-specific financial and insurance use cases. • Built LLM-powered underwriting automation platform — synthetic document generation (bank statements, payslips, credit reports) streamlining financial enterprise assessment workflows. • Collaborated with medical professionals on transformer-based gene-specific medicine discovery. • Led AI development of camera-based automated billing smart cart — real-time object detection eliminating conventional checkout across multiple retail deployments. • Designed CNN-based MobileNet + NIMA facial liveness and recognition system on Kubernetes — 1M+ verifications/month for Jio's enterprise KYC pipeline, 20% accuracy improvement. • Built end-to-end OCR and intelligent document parsing pipeline for ID proof verification, reducing manual review by 60%. • Established MLOps infrastructure with Kubeflow Pipelines and Kubernetes — automated retraining, versioning, CI/CD, zero-downtime deployments for Jio and enterprise clients. • Led 400-person data curation team; built preprocessing and augmentation pipelines accelerating training cycles by 30%.
Sukshi
Apr 20173 yrs 2 mosMay 2020

Hyderabad, Telangana, India

• Retrained and optimised face recognition models for enterprise attendance system — 20% accuracy improvement across diverse demographic datasets. • Built Python/Django + React monitoring platform for real-time model performance benchmarking. • Developed full-featured real-time chat application for e-commerce platform — buyer-seller communication, order tracking, and support workflows.
May 201611 mosMar 2017

Hyderabad, India

• Engineered 3D face reconstruction pipeline using depth camera data for virtual clothing try-on product — bridging hardware sensors, point-cloud processing, and AI productisation. • Installed and supported AI imaging hardware and software systems across multiple retail locations.

Purushotham’s AI Stack

claude
claude
hugging face
hugging face
langchain
langchain
langgraph
langgraph
langsmith
langsmith
Qwen
Qwen

Education

Jawaharlal Nehru Technological University

Bachelor of Engineering - BE, Electrical, Electronics and Communications Engineering

Jawaharlal Nehru Technological University

20124 yrs2016

Secondary school

Good Morning Grammar High School

20046 yrs2010

Join Purushotham and 10k+ other professionals on

Tal Boss

Tal is the app where bosses meet engineers & builders directly, without middlemen

Download Tal Boss

Purushotham Rao Eravalli on Tal

Headline

AI/ML Research Scientist

About

A problem solver & creative thinker of exciting technology problems, I , with my hard work & dedication have very shortly pulled myself from "Zero to One" to a Full Stack & ML engineer to deliver entire bandwidth of technology need. Architecting and administering the model training & performance analysis, I focus on my own "inner learning" too & love nature.

Location

Hyderabad

AI stack

claude

hugging face

langchain

langgraph

langsmith

Qwen

Work experience

Scientist at Jukshio, Jun 2020 - Present · 6 yrs 4 mos, Hyderabad, Telangana, India, • Built production RAG pipelines with hybrid dense-sparse retrieval and LLM grounding for insurance document Q&A, clause extraction, and risk summarisation • Architected multi-agent retail operations platform using CrewAI, LangGraph, and AutoGen — For daily customer interactions, autonomous query resolution, inventory management, and escalation routing. • Fine-tuned LLMs (LLaMA, Qwen, DeepSeek, Mistral) from scratch and with LoRA/QLoRA/PEFT/DeepSpeed for domain-specific financial and insurance use cases. • Built LLM-powered underwriting automation platform — synthetic document generation (bank statements, payslips, credit reports) streamlining financial enterprise assessment workflows. • Collaborated with medical professionals on transformer-based gene-specific medicine discovery. • Led AI development of camera-based automated billing smart cart — real-time object detection eliminating conventional checkout across multiple retail deployments. • Designed CNN-based MobileNet + NIMA facial liveness and recognition system on Kubernetes — 1M+ verifications/month for Jio's enterprise KYC pipeline, 20% accuracy improvement. • Built end-to-end OCR and intelligent document parsing pipeline for ID proof verification, reducing manual review by 60%. • Established MLOps infrastructure with Kubeflow Pipelines and Kubernetes — automated retraining, versioning, CI/CD, zero-downtime deployments for Jio and enterprise clients. • Led 400-person data curation team; built preprocessing and augmentation pipelines accelerating training cycles by 30%.

Research And Development Engineer at Sukshi, Apr 2017 - May 2020 · 3 yrs 2 mos, Hyderabad, Telangana, India, • Retrained and optimised face recognition models for enterprise attendance system — 20% accuracy improvement across diverse demographic datasets. • Built Python/Django + React monitoring platform for real-time model performance benchmarking. • Developed full-featured real-time chat application for e-commerce platform — buyer-seller communication, order tracking, and support workflows.

Field Engineer at TRUPIK VIRTUALIZATION PRIVATE LIMITED, May 2016 - Mar 2017 · 11 mos, Hyderabad, India, • Engineered 3D face reconstruction pipeline using depth camera data for virtual clothing try-on product — bridging hardware sensors, point-cloud processing, and AI productisation. • Installed and supported AI imaging hardware and software systems across multiple retail locations.

Education

Bachelor of Engineering - BE, Electrical, Electronics and Communications Engineering, Jawaharlal Nehru Technological University

Secondary school, Good Morning Grammar High School