Industry Experience

Astha.IT
Astha.IT Current
Nov. 2025 - Present
AI Engineer II
Dhaka, Bangladesh
Product: TalentX
  • Built a Job Builder Agent that combines document parsing, field-level extraction, and human-in-the-loop refinement to transform user requirements or PDF/DOCX inputs into structured, editable job specifications.
  • Designed a Profile Screening Agent that performs resume information extraction, career-page parsing, semantic candidate–job matching, and ranking to identify qualified candidates against job-specific requirements.
  • Developed a Job Fit Scoring Agent that dynamically generates evaluation criteria and weights from job requirements, applies weighted scoring and requirement-level candidate assessment, and sends HRs detailed evaluation reports with interpretable scores and evidence-based explanations via email.
  • Architected a Talent Matching Agent that converts HR queries into structured retrieval constraints and combines semantic vector search, metadata filtering, query optimization, and relevance reranking to retrieve evidence-backed candidate recommendations, with dynamic removal of outdated profiles and ingestion and updating of new profiles.
  • Optimized large-scale CV processing through batch execution, parallel processing, and redundant LLM-call reduction, enabling the evaluation of 800–1,000 candidate profiles and cutting manual effort from 5 days to under 12 hours.
Tech Stack PythonLangGraphLangChainOpenAI (GPT-4o, text-embedding-3-large)Google Cloud StorageGoogle App EngineGoogle Compute EnginePostgreSQLAmazon SES (SMTP)FastAPIReactTypeScript
Dexian (Bangladesh) Limited
Dexian (Bangladesh) Limited
Jul. 2025 - Oct. 2025
Senior Application Developer
Dhaka, Bangladesh
Project: ShareFlow Agent
  • Architected a multi-agent RAG system integrating SharePoint and web sources via dynamic tool invocation, leveraging ReAct reasoning and source-aware context construction for knowledge-grounded responses.
  • Developed a Retrieval Agent with ReAct planning, query reformulation, retrieval filtering, and custom guardrails to constrain responses to internal documents and reduce unsupported generation.
  • Orchestrated an OCR Agent to extract content from diverse unstructured sources (scanned PDFs, images, and DOCX files) with rate limiting to support fault-tolerant performance.
  • Integrated a Web Browser Agent with controlled web retrieval and instruction-aware source selection to augment knowledge-grounded responses with current external information when enterprise sources were insufficient.
  • Implemented dynamic knowledge-base management with incremental SharePoint synchronization, file-level tracking, and memory updates, allowing users to add or remove knowledge sources without rebuilding the complete index.
  • Designed session-aware conversational workflows with persistent history, context management, and session reset mechanisms, and integrated the system with Microsoft Bot Framework for deployment within Microsoft Teams.
  • Directed system architecture and mentored junior developers through design reviews, debugging, code reviews, and technical knowledge-sharing, achieving an average 60–65% annual operational cost reduction through agent and inference optimization for 80 Sales Managers with 50+ interactions per day.
Tech Stack PythonLlamaIndexAzure OpenAI (GPT-4.1, GPT-4o, text-embedding-3-large)Azure Bot ServicesAzure SQL DatabaseAlloyDB for PostgreSQL (pgvector)Azure App ServiceReactFastAPIRagas
Dexian (Bangladesh) Limited
Dexian (Bangladesh) Limited
May 2024 - Jun. 2025
Application Developer
Dhaka, Bangladesh
Project: OrgInfo
  • Implemented an OrgChart Extraction Agent with Tree-of-Thought (ToT) reasoning, visual parsing, hierarchical inference, and structured entity mapping to reconstruct organizational hierarchies from organograms and map extracted roles to corresponding Bullhorn records in a relational database.
  • Developed an OrgQuery Agent that converts natural language queries into optimized SQL (text-to-SQL) using Chain-of-Thought (CoT) and self-consistency prompting, retrieves organizational hierarchy from the relational database, and presents results through the OrgChart interface for hierarchical visualization.
  • Engineered OrgInfo Assistant to allow users to interact with specific organizational hierarchies by converting queries to SQL, executing them on a temporary organization-specific database to reduce load on the primary database, and converting the retrieved results to natural language.
  • Created an OrgActivity Summarization Agent to generate 7-day summaries of organizational activities by extracting relevant notes on placements, submissions, and communication logs from the Bullhorn database.
  • Automated Bullhorn synchronization through scheduled data ingestion and incremental database updates, cutting organizational hierarchy search time by 75% by eliminating repeated direct queries against the source system.
Tech Stack PythonLangChainLangGraphAzure OpenAI (GPT-4.1, GPT-4o mini, text-embedding-3-small)Azure SQL DatabaseAzure App ServiceOpenCVReactFastAPIWebSocket
Project: RFPMatcher
  • Architected a document-centric RAG pipeline with Chain-of-Thought reasoning and 12 specialized analytical modules to extract key information (e.g., project scope and deliverables) from Request for Proposal (RFP) documents.
  • Orchestrated a master RFP database by processing historical responses with metadata (generated answers, summaries, win/loss labels, and question categories) and vector embeddings to support semantic search.
  • Designed a Past Experience Matcher system that analyzes new RFP documents by extracting key client requirements, generating binary (Yes/No) and descriptive questions for each section, and querying a master database to identify relevant past experiences through semantic similarity matching.
  • Developed a Table of Contents (TOC) generation workflow that analyzes extracted RFP key information against predefined section libraries (10 standard and 12 non-standard sections) to intelligently select and prioritize relevant sections based on project requirements.
  • Implemented a section-based Proposal Writing Assistant that generates initial content for each TOC section using extracted RFP information and relevant past experience, and enables iterative refinement through targeted chat interactions while maintaining conversation history.
  • Accelerated decision-making through predictive insights, with manual review of 100+ pages reduced by 3–5 days and Proposal Managers able to focus on strategic bid development.
Tech Stack PythonLlamaIndexAzure OpenAI (GPT-3.5 Turbo, GPT-4, text-embedding-3-large)Prompt EngineeringAlloyDB for PostgreSQL (pgvector)CouchDBAzure App ServiceReactFastAPIDeepEvalDocker
Project: AgentDexi
  • Formulated a RAG system to identify technological trends and track the top 20 most in-demand skills by extracting information from job descriptions scraped from external company postings.
  • Orchestrated a web Research Agent that autonomously performs web searches using the Google Search API and Tavily Search API to extract company information from diverse sources, retrieving the top 10 ranked URLs with contextual preview summaries and filtering results based on relevance to support downstream data extraction and analysis tasks.
  • Implemented a Website Scraping Agent utilizing ScraperAPI to extract targeted data from discovered URLs, processing raw HTML/text content into structured data, and built a custom RAG pipeline that filters contextually relevant information and generates structured outputs, including tables, organizational reports, and business intelligence summaries.
  • Built a Transcript Agentic RAG pipeline that ingests IT/tech video URLs, processes transcripts through semantic chunking, and produces structured, context-aware summaries to surface emerging technologies and in-demand skills.
  • Designed an automated data analysis pipeline with interactive graphs and charts that surface company-wise hiring patterns and role-specific demand, equipping 250+ Technical Recruiters with competitive insights to make informed, data-driven talent acquisition decisions.
Tech Stack PythonLangChainLangfuseCrewAIAzure OpenAI (GPT-4o, text-embedding-3-small)ChromaDBJobSpyReactFastAPI
Project: CaseAligner
  • Designed and deployed an LLM application that repurposes existing client-facing PowerPoint presentations for case studies, with chain-of-thought prompting to transform them into new practice areas and industries, enabling rapid generation of domain-specific demo presentations.
  • Implemented an interactive chat interface linked to separated slides, allowing users to query and modify specific slide content, with session-wise conversation history stored for reference.
  • Created a summarization feature for all existing case studies, highlighting technology stacks, work types, key benefits, and outcomes to assist users in quickly identifying relevant content.
  • Developed comprehensive search functionality to locate information across all generated case studies, including references to specific slide numbers for precise navigation.
  • Added export functionality to download slides in the company's official presentation template.
  • Built a user interface for users to download and edit the knowledge base of case studies used for generating new ones.
  • Significantly accelerated demo preparation by saving 1-2 full days of slide crafting, allowing Salespersons to focus more on client engagement and closing deals.
Tech Stack PythonLlamaIndexAzure OpenAI (GPT-3.5 Turbo)ReactFastAPIAzure App ServiceDocker

Reviewer Experience

FinNLP-2026 @ EMNLP-2026 Scientific Reports Language Resources and Evaluation Discover Mental Health Cluster Computing

Tutoring Experience

Private Academic Tutor
2020 - 2023
Mathematics, Physics & Chemistry
Dhaka, Bangladesh
  • Provided personalized one-on-one tutoring in Mathematics, Physics, and Chemistry for high school students.
  • Developed customized lesson plans and practice exercises, monitored academic progress, and conducted regular assessments with feedback.