Help & Supportcareers@gravityer.com
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
AI Engineer with less than a year in AI/ML, LLMs & Cloud Deployment
AI Engineer with a background in Computer Science and Engineering, specializing in Artificial Intelligence and Machine Learning. Experienced in designing, developing, and deploying AI-powered applications with hands-on industry expertise in LLMs, Generative AI, and workflow automation. Skilled in Python, RAG pipelines, vector databases, and cloud-based deployment, with a proven ability to bridge academic knowledge and real-world applications. Passionate about building scalable, reliable AI solutions and guiding teams to leverage emerging technologies for impactful results.
Daffodil International University
B.Sc. · Computer Science and Engineering
August 1, 2021 – June 30, 2025
Betopia Group - Softvence Omega
AI Developer
December 1, 2025 – Present
Dhaka, Dhaka Division, Bangladesh
Real-Time AI Voice Receptionist with tool calling
January 1, 2025 – Present
Developed a real-time, low-latency AI voice receptionist agent capable of bidirectional conversations with callers. Engineered a FastAPI microservice backend integrating Twilio, ElevenLabs, and whisper for seamless interactions. Used OpenAI for real-time chat with dynamic context and TTL-based session memory.
View ProjectSpeaklerRag: Low-Latency Hybrid Retrieval and Coreference Resolution Pipelines in Multimodal AI Systems
January 1, 2025 – Present
Developed a Bengali RAG system for contextual document querying and AI-generated responses. Built semantic retrieval pipelines using embeddings, vector search, and LLMs. Designed FastAPI backend APIs for real-time document ingestion and querying. Improved retrieval accuracy with multilingual processing and context-aware search.
View ProjectProductImageGen: Automated Multi-Modal Conditioned Image Generation and Pipeline Orchestration Framework
January 1, 2025 – Present
Designed and implemented an asynchronous FastAPI backend to orchestrate complex vision synthesis tasks conditioned on multi-attribute inputs (brand identity, product metadata, and structured briefs). Developed a modular pipeline integrating image generation clients, programmatic background removal stubs, and metadata schema validations to ensure deterministic, production-grade outputs. Engineered a scalable job-polling architecture leveraging asynchronous task management to decouple compute-heavy generation requests from HTTP client responsiveness.
View ProjectSpatiotemporal Feature Extraction and Textual Grounding in Video Analytics
January 1, 2025 – Present
Engineered a comprehensive spatiotemporal video analysis pipeline utilizing OpenCV Farneback optical flow for continuous motion tracking and pixel-differential modeling for automated shot-cut detection. Integrated multi-modal detection capabilities by combining YOLOv8 architectures for spatial object-person dominance evaluation alongside Tesseract OCR for text extraction and semantic profiling. Designed a modular, enterprise-grade Python package architecture configured via YAML, featuring rigorous integration testing suites (Pytest) and localized CLI execution for reproducible video preprocessing workflows.
View ProjectParameter-Efficient Fine-Tuning of Autoregressive Language Models for Low-Resource Empathetic Dialogue Generation
January 1, 2025 – Present
Designed and executed an end-to-end fine-tuning pipeline to adapt the Meta LLaMA 3.1 8B Instruct model to the Bengali language domain utilizing the Bengali Empathetic Conversations Corpus. Optimized computational efficiency by implementing 4-bit quantized base model loading coupled with Parameter-Efficient Fine-Tuning (PEFT/LoRA) via Unsloth, significantly minimizing VRAM constraints during training. Developed automated text preprocessing modules using Hugging Face chat templates alongside robust validation and evaluation infrastructure measuring token perplexity, BLEU, and ROUGE-1/2/L scores.
VehicleInsurance: A Modular and Containerized MLOps Pipeline for Scalable Risk-Prediction Architecture
January 1, 2025 – Present
Engineered a production-grade MLOps pipeline featuring modular components for automated data ingestion, validation, transforming, training, and evaluation. Implemented structured schema validations via configuration files to enforce data integrity and prevent drift before model training. Integrated cloud and database adapters leveraging AWS S3 for model tracking and MongoDB for scalable unstructured data storage.
View ProjectOMRChecker: Automated Document Alignment and Spatial Analysis Framework for Optical Mark Recognition
January 1, 2025 – Present
Designed a classical vision pipeline utilizing Gaussian blur, Canny edge detection, and contour analysis to isolate document boundaries. Implemented geometric rectification algorithms leveraging homography matrices and perspective transformations to standardize coordinate planes. Engineered pixel-density segmentation modules using NumPy to partition target document regions for threshold-based evaluation.
View ProjectData Science Math Skills
Coursera
June 1, 2026 – Present
Cultural Fit Analysis
The candidate's diverse range of personal projects, covering areas like real-time voice AI, multimodal RAG, image generation, video analytics, and MLOps pipelines, demonstrates a strong passion for AI and a proactive learning attitude. Participation in volunteer activities like 'AIML Professional Community Bangladesh' and 'NLP Club at DIU' indicates a collaborative spirit and engagement with the AI community. The breadth of technologies used across projects (OpenAI, LLMs, computer vision, MLOps tools, cloud platforms) suggests adaptability and a willingness to explore new tools, which aligns well with a dynamic AI engineering environment. The candidate's target role as an AI Engineer is well-aligned with their project experience and technical skills.
Soft Skills & Operational Fit
The candidate's project descriptions indicate a strong ability to design and implement complex systems, suggesting good problem-solving and architectural thinking. The emphasis on modularity, testing (Pytest), and containerization (Docker) points to an operational mindset focused on reproducibility and maintainability. The description of managing end-to-end model lifecycles further supports a structured and responsible approach to development. However, without direct interview data, specific soft skills like teamwork, leadership, or adaptability cannot be fully assessed.