Open to opportunities

pedram esmaeili

@pedramesmaeili

AI Software Engineer with 16+ years building production-grade, scalable systems and integrating advanced AI/ML into real products.

Turkey

Message

What I'm looking for

I’m looking to build production-grade AI systems that solve real-world problems—combining backend engineering, RAG, and real-time/multimodal experiences—while having room to lead delivery and continuously optimize performance, reliability, and deployment.

I’m an AI Software Engineer with 16+ years of experience designing, building, and scaling robust software systems across the full development lifecycle. I lead delivery of high-performance, production-grade solutions by combining backend engineering with modern AI/ML frameworks.

I’ve built AI systems that perform at scale—like a food categorization engine achieving >99% accuracy across millions of menu items, and a conversational AI that routes between a FAQ engine and RAG with <2s end-to-end latency and >98% production answer accuracy. I’ve also launched secure on-prem AI tooling using Open WebUI, enabling large teams to work without cloud data exposure.

I’m especially energized by real-time and multimodal applications, from sub-second GPT-4o voice and avatar experiences to RAG layers that reduce token usage by 40% and latency by 30%. My work blends data pipelines, optimization, and deployment discipline to turn complex AI into dependable products.

Experience

Work history, roles, and key accomplishments

Current

AI Software Engineer

Current

Tapsi Food

Oct 2025 - Present (8 months)

Architected a food categorization system combining prompt engineering with a domain-tuned ParsBERT model, achieving >99% classification accuracy across million-scale production menu items. Built production conversational AI with LangGraph and a Qdrant-backed RAG pipeline, delivering <2s end-to-end latency and >98% answer accuracy.

Prompt Engineering ParsBERT Fine Tuning LangGraph Retrieval Augmented Generation (RAG)Qdrant fastAPI WebSockets

AI Software Engineer

Dilz Estate

Jan 2025 - Sep 2025 (8 months)

Achieved sub-second real-time responses (<1s) using OpenAI GPT-4o Real-time API with persistent WebSocket connections and optimized session management. Built multimodal conversational experiences with voice synthesis and AI avatars (5+ languages), achieving <1.5s voice generation and <3s avatar rendering latency.

WebSockets Function Calling Retrieval Augmented Generation (RAG)Vector Databases OpenAI API Speech Synthesis

AI Software Engineer

Teammateapp

Sep 2024 - Jan 2025 (4 months)

Refactored a full-stack CRM (NestJS + Angular) with a modular backend architecture, improving scalability and maintainability by 40%. Led R&D for a RAG-based AI assistant using help documentation and delivered CRM features including “WorkFlow” task automation while resolving 40+ critical front- and back-end issues to reduce downtime incidents.

NestJS Angular Retrieval Augmented Generation (RAG)backend architecture API Integration Workflow Automation

AI Software Engineer

Celeste AI

Jan 2024 - Aug 2024 (7 months)

Developed a conversational AI that converts natural-language chat into optimized SQL queries, improving data accessibility for non-technical users. Built and deployed an NLP pipeline using GPT/T5/BERT/Transformer models and a NestJS REST API, delivering an 80% improvement in SQL recognition accuracy.

BERT Transformer Models SQL Query Generation NestJS TensorFlow PyTorch

Senior Back-End Engineer

Autonomous Professional Development

Jun 2021 - Nov 2023 (2 years 5 months)

Designed an AI-powered matching algorithm for a talent platform using cosine similarity, fuzzy logic, and machine learning to improve candidate-to-job recommendations. Migrated REST APIs to a microservices architecture, reducing response times by 30% and improving scalability while delivering a T5-based sequence-to-sequence model with 0.01 error rate.

Microservices Architecture RESTful APIs Cosine Similarity Machine Learning T5 Sequence To Sequence Models API Refactoring

Full Stack Software Engineer

Praxis Payment

Mar 2018 - Jun 2021 (3 years 3 months)

Developed and deployed three Python-based AI algorithms for fraud detection and risk management in a cryptocurrency wallet management system supporting 10,000+ wallets. Built and enhanced a secure blockchain-based platform with Node.js and PostgreSQL, improving backend performance with caching and database optimization to reduce response times by 30%, supporting 10,000 successful transactions dail

Python Fraud Detection Node PostgreSQL Caching Database Optimization Crypto Wallets System Reliability

Full Stack Software Engineer

Tarhan Mandegar

Jun 2013 - Feb 2018 (4 years 8 months)

Led a team delivering large-dataset projects using scalable architecture and agile methodology. Built an e-wallet and credit system with AI-driven fraud detection and user profile analysis, and oversaw healthcare, e-commerce, and OCR signature recognition initiatives including an XRAY breast-cancer detection pilot.

Team Leadership Scalability Agile Methodology E Wallet & Credit Systems Fraud Detection Healthcare Applications

Back-End Developer / AI

Pump Iran

Feb 2011 - Jun 2013 (2 years 4 months)

Built and implemented a machine-learning pump selection algorithm using MATLAB, Python, and ASP.NET/C# that increased operational efficiency by 50%. Revamped backend architecture and applied advanced query optimization techniques, reducing application load time by 40%.

Machine Learning Matlab Python ASP.NET C#Query Optimization backend architecture Data Processing

Full Stack Developer

Negar Hack

Feb 2008 - Feb 2011 (3 years)

Built robust back-end systems and REST APIs for e-commerce and financial management clients, incorporating AI-based algorithms where applicable. Delivered scalable solutions using HTML, CSS, ASP.NET/C#, and SQL Server that met or exceeded customer expectations.

RESTful APIs ASP.NET C#SQL Server HTML CSS Algorithms Financial Systems Scalability