Leela Akula

Machine learning Developer

I am passionate about crafting intelligent, production-ready systems that transform raw data into actionable insight.

Machine Learning engineer at Zotok

MLLanggraph & LangChainLlamaPythonAzureAWSprompt engineeringTypeScript

Machine Learning

  • Model Architectures
  • Performance Optimization
  • intelligent retrieval Mechanisms.

Backend Development

  • API Design & Development
  • Database Architecture
  • Real-time Systems

Prompt Engineering

  • Designing effective prompts for LLMs
  • Contextual input formatting
  • Tools: LangChain, PromptFlow, PromptLayer, Guidance

Machine Learning Projects

Seller-buyer chatbot Platform

A high-performance AI-CHATBOT solution handling 100K+ daily queries with real-time inventory and ML-powered recommendations using langgraph.

Frontend Architecture

  • • Streamlit
  • • JS
  • • Css

Backend Systems

  • • Langgraph
  • • Python
  • • Tools
  • • Azure models
  • • OpenAI query Search

Key Achievements

  • • 99.99% Uptime Deployment
  • • 200ms Average API Response Time
  • • 80% Reduction in human interaction

System Architecture

streamlit Frontend ( Client Components)supervisor Servicesupervisor ServicetoolsPromptAzure modelsOPENAIsearch Search

Real-time Analytics Platform

A real-time intelligent parsing system leveraging GPT-4.0-mini via PromptFlow for seamless data extraction across product descriptions, images, invoices, and payments.

BackEnd Features

  • • promptFlow
  • •Azure model
  • • Yaml files
  • • Fast API

Performance Metrics

  • • 1M+ Events/Minute Processing
  • • Sub-second Query Response
  • • 99.99% Data Accuracy

Data Flow Architecture

FastAPILLM: GPT-4.0-miniPromptFlowfind the type of fileextract the textidentify the file contextget prompt instructionsfind the keysreturn the values

Product Search System

A scalable ML-powered search engine that transforms raw product names into high-dimensional embeddings using text-embedding-ada, stores them efficiently, and retrieves relevant matches using vector similarity with Azure Cognitive Search and FAISS.

BackEnd Features

  • • Text Embedding with text-embedding-ada
  • • Azure Cognitive Search Integration
  • • FAISS Indexing
  • • File I/O with Dynamic Embedding Generation

Performance Metrics

  • • 98%+ Semantic Search Precision
  • • Sub-second Query Response
  • • Sub-300ms Retrieval Latency

Data Flow Architecture

Input Files3event triggerGenerate EmbeddingsFAISS + CognitiveProduct QuerySimilarity Searchk=3Response to API

Let's Build Something Amazing

Looking for a Machine Learning developer who can architect and implement complete solutions? Let's discuss your project.

Hyderabad