Hi, I'm

Adari Yashwanth

AI / ML Engineer Backend Engineer Founding Engineer

I build production-scale AI systems, distributed backends, and LLM-powered products. IIT Guwahati alumnus. Kaggle Competitions Expert.

View Work Get in Touch
Adari Yashwanth
3+ Years Industry Experience
1B+ Records Processed
Top 1.4% Kaggle Competition
2 Research Publications

Experience

AagmanAI

Founding Engineer

Jul 2025 – Apr 2026 Remote
  • Architected a stateful plan-act-observe agent framework using Playwright + LLM tool calling, improving success rate on complex workflows by 22% through deterministic retries and fallback policies.
  • Architected a distributed ingestion platform (Node.js, AWS) processing 1B+ records with real-time tick ingestion, ClickHouse for analytics, PostgreSQL for transactional workloads, and S3 for durable storage.
  • Built a multi-stage LLM inference system using QLoRA fine-tuned Gemma 2B, achieving 91% accuracy, 36% P95 latency reduction, and ~28% cost savings at scale.
  • Built production-scale RAG and retrieval systems over millions of news articles and SEC documents with optimized chunking and ranking.
  • Designed a multi-tenant data platform with RBAC and a 10M+ entity relationship graph via entity resolution across heterogeneous financial datasets.
  • Designed low-latency backend services improving throughput by 3.2× via async execution and optimized task scheduling.
Node.jsLLMClickHousePostgreSQLAWSPlaywrightRAG

Blinq.io

Algorithmic / Founding Engineer

Aug 2024 – Jun 2025 Remote
  • Designed and optimized LLM-based agent workflows, improving system accuracy by 16% through better prompt chaining, evaluation, and tool usage.
  • Built a FastAPI automation service integrating LLM APIs for Jira workflows, reducing manual effort by 30%.
  • Fine-tuned Qwen2-VL and Qwen2.5-VL multimodal LLMs for browser automation, improving task accuracy from 66% to 71.4%.
  • Led authentication and access control using Descope SSO + RBAC, securing 3K+ users with universal middleware.
  • Built distributed storage pipeline (MongoDB, Kafka, Docker) processing 3K+ reports/month and a git-workspace service for 10K+ tests.
  • Deployed scalable inference pipelines on AWS SageMaker with monitoring, logging, and autoscaling.
FastAPILangChainSageMakerQwen2-VLKafkaMongoDBReact

Edgeverve Systems Limited

Member of Technical Staff

Sept 2023 – Aug 2024 Bangalore
  • Built an AI copilot platform using RAG + fine-tuned LLMs, improving enterprise workflow efficiency by 25%.
  • Designed high-performance search over millions of records using OpenSearch, PGVector, ChromaDB, reducing query latency by ~40%.
  • Developed ML solutions for address matching (dense embeddings + BERT) and speech-to-text optimized for Indian accents.
  • Improved evaluation pipelines, increasing macro F1 score by 30% through better dataset curation and metric design.
  • Deployed production systems on Azure with monitoring, CI/CD, and performance tuning.
RAGOpenSearchPGVectorBERTAzureCI/CD

Udgam, IIT Guwahati

Web Operations Lead / Developer

Jan 2020 – Sept 2021 Guwahati
  • Led a team of 5 engineers building high-traffic event platforms supporting 10K+ users with zero downtime during peak festival traffic.
  • Developed full-stack systems using React, Django, PostgreSQL, and Firestore with secure, PCI-compliant payment integration.
ReactDjangoPostgreSQLFirestore

Technical Skills

Languages

TypeScriptPythonJavaScriptGoJavaC++C

Backend & Web

Node.jsFastAPIDjangoFlaskReactREST APIs

Databases

PostgreSQLMongoDBClickHouseOpenSearchRedisPGVectorChromaDB

Cloud & DevOps

AWSGCPAzureDockerKafkaTerraformAWS CDKCI/CD

AI / ML

LLM IntegrationRAG PipelinesFine-TuningSageMakerLangChainEmbeddingsAgent Frameworks

Systems

Distributed SystemsAsync ExecutionLinuxBashGit

Research & Publications

First Author · Lead Analyst

Some important observations from microstructural analysis of the quartz-rich mylonites of the Main Central thrust from Sikkim, eastern Himalaya

Deformation Mechanisms, Processes and Fabric Analysis in Earth Materials from the Global to the Nano Scale · IIT Kharagpur · Feb 2024

Led all quantitative analysis and mathematical modeling including Gaussian mixture modeling, vorticity axis analysis, and microstructural parameter extraction.

View on ResearchGate →
Co-Author · Data Analysis Lead

Tracking the research trends in geodynamic evolution of the Himalayan metamorphic core: The past, the present, and the future

SERB-DFG Week of Young Researcher · Wadia Institute of Himalayan Geology, Dehradun · Nov 2023

Responsible for data analysis and mathematical interpretation for strain and vorticity axis analysis.

View on ResearchGate →

Kaggle

Silver
Competitions Expert

Automated Essay Scoring 2.0

🥈 38th place out of 2,706 teams · Top 1.4%

  • Built high-performance models using DeBERTa-based transformers with feature augmentation from TF-IDF, CountVectorizer, and spaCy linguistic features.
  • Optimized for Quadratic Weighted Kappa (QWK), achieving a score of 0.835 via ensembling and robust cross-validation strategies.
View Kaggle Profile →

Education

Indian Institute of Technology (IIT), Guwahati

B.Tech · Jul 2019 – Jul 2023

Let's Build Something

Open to founding engineer roles, AI/backend positions, and interesting problems worth solving.

y9yashwanth@gmail.com linkedin.com/in/adariyash +91 76378 34925