Help & Supportcareers@gravityer.com
AI Research Engineer with 5+ years in Computer Vision, NLP & LLMs
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
Highly knowledgeable and skilled Al engineer seeking full-time opportunities in Al engineering or a related field. Proficient in developing popular computer vision models such as GANs, diffusion models and ViTs using popular libraries like PyTorch and Tensorflow. Possess advanced capabilities in using open-source LLM orchestration libraries like LangChain to develop RAG and other NLP applications. Proven ability to manage time effectively and deliver under pressure, evident from handling project work alongside a teaching role at Brunel University. Possess strong programming skills in Python, Java and C++ acquired through professional work experience and personal projects.
Brunel University
PhD · Computer Science
October 1, 2020 – October 1, 2024
Brunel University
BSc · Computer Science (AI)
September 1, 2017 – July 1, 2020
Metropolitan Police
Data Scientist
February 1, 2024 – Present
India
Brunel University
System Administrator
January 1, 2022 – September 1, 2022
India
Brunel University
Senior Research Assistant
December 1, 2021 – September 1, 2022
India
Brunel University
Doctoral Researcher
September 1, 2020 – November 1, 2023
India
Brunel University
Graduate Teaching Assistant
September 1, 2020 – November 1, 2023
India
Triaging FOI request with LLM
June 19, 2026 – Present
Processed and analysed a high volume of daily FOI requests received by the police organisation, identifying a significant number of non-legitimate submissions. Utilised historical records containing labelled datasets of valid and invalid FOI requests to develop a classification model. Fine-tuned a GPT-2 model on the historical dataset to automate the classification of FOI requests as valid or invalid. Improved efficiency for public relations and data rights teams by enabling them to focus solely on legitimate FOI requests, optimising workflow and resource allocation.
View ProjectFirearm Detection
June 19, 2026 – Present
Utilised YOLOv3 to effectively identify and draw bounding boxes around firearms in live-stream video footage, enhancing real-time threat detection capabilities. Conducted extensive experimentation with both Python and C++ to compare computational efficiency and resource demands. Successfully productionised the model using Docker to ensure smooth deployment on Azure. Leveraged Gradio to create user-friendly interfaces, deploying the model as Databricks apps.
View ProjectSupporting intelligence gathering with GraphRAG
June 19, 2026 – Present
Developed a GenAI chatbot to streamline intelligence gathering for officers overwhelmed by numerous internal policies. Designed a solution that interprets officers' initial descriptions and retrieves relevant policy guidance. Utilised a GraphRAG approach: transformed internal policy documents into a Neo4j-based knowledge graph. Employed LangChain's GraphRAG library to generate Cypher queries using the LLaMA model, enabling precise extraction of relevant policy nodes. Chatbot provides real-time instructions and recommendations aligned with internal protocols, improving operational efficiency and compliance.
Fleet Simulation
June 19, 2026 – Present
Developed a discrete event simulation (DES) model to simulate the movement of fleet vehicles. Enabled the fleet management team to experiment with resources such as the number of vehicles or driver officers. Demonstrated how long it takes a unit to arrive at the scene. Illustrated potential impacts of insufficient vehicles or police drivers on arrival times. Simulated the number of incidents occurring in a 24-hour period, reflecting low incidents in the morning and high incidents during the day/evening.
Image-based Virtual Try-On
June 19, 2026 – Present
Developed a novel approach utilising Generative Adversarial Network (GAN) to replace a person's garment with a desired item. Contributions include refining the candidate representation to incorporate enhanced posture information, implementing a truncated U-Net for improved performance, and utilising Affine Transform for efficient garment transformation. Evaluated the performance of the developed model by comparing it to state-of-the-art models in the field. The results showcased superior qualitative and quantitative outcomes, highlighting the model's ability to generate realistic and high-quality try-on images.
View ProjectMulti-Pose Virtual Try-On
June 19, 2026 – Present
Incorporated a unique feature into the traditional virtual try-on model, enabling the ability to change the posture of the person while simultaneously replacing the garment. Contributed to the development of a pipeline that integrates the pose transfer module, based on a StyleGAN architecture, with a traditional virtual try-on framework primarily utilising U-Net. Demonstrated the superiority of the model by synthesising images of significantly higher quality compared to previous works. The generated images reached a supreme level of realism, surpassing the limitations of prior research efforts.
View ProjectImage-to-Video Synthesis
June 19, 2026 – Present
Developed a generative model for synthesising videos performing random yet plausible movements from fashion images. Leverages latent video diffusion techniques, comprising pseudo-3D convolutional layers designed to optimise efficiency and ease of training, a cross-attention mechanism that effectively incorporates the fashion image into the video generation process whilst being accompanied by an adapter module to improve the efficacy. This model significantly outperformed our competitors by producing smoother and better-conditioned videos. It enhances the potential shopping experience.
View ProjectFinal Year Project
June 19, 2026 – Present
Developed a Python solution that utilises pose estimation to recognise instances of physical crimes in CCTV footage. Created a rule-based AI system capable of identifying physical contact between individuals and recalling the location of the attack. Analysed complex problems and effectively broke them down into smaller sections, facilitating the efficient development of the software.
View ProjectMulti-player Yahtzee game
June 19, 2026 – Present
Developed a multiplayer Yahtzee game in Java that enables 3 or more clients to play together. Implemented multithreading on the server side to manage and synchronise multiple connected clients. Designed and developed the server-client communication protocol, ensuring reliable data transmission and handling potential network issues. Implemented game mechanics such as rolling dice, scoring, and tracking player turns, ensuring accurate gameplay according to the Yahtzee rules. Utilised concurrent control to handle scoreboard requests and updates from clients, preventing multiple client threads from modifying the scoreboard simultaneously.
View ProjectGroup Project - Android Development
June 19, 2026 – Present
Contributed to a university team project with six members, collaborating on the development of an Android dating app. Actively participated in brainstorming sessions and feature planning meetings, offering creative ideas and suggestions to enhance the app's functionality and user experience. Worked closely with team members to assign tasks, set milestones, and establish a collaborative development approach, ensuring efficient progress and timely completion of the project. Effectively communicated with team members to establish a collaborative development approach and fostered trust, resulting in an enjoyable project completion experience. Took ownership of specific modules within the app, developing clean and well-structured code while adhering to best practices and coding standards. Conducted thorough testing when merging code with other team members, working together to ensure the app was functioning properly and free of conflicts. Self-taught Android Programming through online resources, dedicating extensive hours to research and enhancing problem-solving skills for creating high-quality Android apps.
Image-based virtual try-on: Fidelity and simplification
Signal Processing: Image Communication
January 1, 2024 – Present
Transforming Digital Marketing with Generative AI
Computers
January 1, 2024 – Present
Dynamic Fashion Video Synthesis from Static Imagery
Future Internet
January 1, 2024 – Present
StyleVTON: A multi-pose virtual try-on with identity and clothing detail preservation
Neurocomputing
January 1, 2024 – Present
Deep Learning in Virtual Try-On: A Comprehensive Survey
IEEE Access
January 1, 2024 – Present
Svton: Simplified virtual try-on
Computer Science Brunel PhD Symposium
May 1, 2022 – Present
SVTON: Simplified Virtual Try-On
International Conference on Machine learning and Applications
January 1, 2022 – Present
CALMS: Modelling the long-term health and economic impact of Covid-19 using agent-based simulation
Plos one
January 1, 2022 – Present
FACS-CHARM: a hybrid agent-based and discrete-event simulation approach for Covid-19 management at regional level
2022 Winter Simulation Conference (WSC)
January 1, 2022 – Present
Deep Learning for Personalised Marketing
Computer Science Brunel PhD Symposium
May 1, 2021 – Present
Cultural Fit Analysis
The candidate's diverse project portfolio, ranging from academic research in virtual try-on and video synthesis to professional applications in law enforcement (firearm detection, FOI triaging, intelligence gathering), demonstrates adaptability and a broad interest in applying AI to various domains. Their PhD research and multiple publications indicate a strong commitment to continuous learning and contributing to the scientific community, which aligns well with a research-oriented role. The experience in a university setting as a System Administrator and Teaching Assistant also shows a collaborative and supportive mindset. The blend of academic rigor and practical application in their current Data Scientist role suggests a well-rounded individual who can bridge theoretical knowledge with real-world impact.
Soft Skills & Operational Fit
The candidate demonstrates strong problem-solving skills, evidenced by breaking down complex problems in their Final Year Project and self-teaching Android programming. Their experience as a Graduate Teaching Assistant and Senior Research Assistant highlights communication and collaboration skills. The ability to manage project work alongside a teaching role indicates effective time management and ability to deliver under pressure. Participation in a group Android project also shows teamwork and communication. The candidate's interests in Karate and gym suggest discipline, focus, and stress handling capabilities.