Job Purpose
- We are seeking a Freelance Research Engineer to provide end-to-end experimental implementation support for an advanced AI research project focused on chart-grounded fact correction.
- In this role, your objective will be to work with chart images and associated textual claims to verify whether each claim is factually consistent with the visual data. Where inconsistencies are detected, you will implement systems to generate an accurate, corrected claim grounded in the visual evidence. The dataset and verification-labeling phases are already completed; your immediate priority will be coding, running baseline experiments, implementing the proposed framework, conducting comprehensive evaluations, and providing paper-writing support for a target Q1/Q2 journal.
- Fact verification and correction: claim-evidence alignment, inconsistency detection, corrective text generation
- Chart understanding: bar charts, line charts, pie charts – value extraction, trend analysis, comparative reasoning
- Employment Type: Freelance / Contract (Fully Remote)
- Primary Focus: Deep Learning engineering, VLM fine-tuning, evaluation framework design, and academic reporting.
- Target Publication: Q1/Q2 Tier Journal.
- Required Qualifications & Experience
- Education: Master’s degree or PhD in Computer Science, Artificial Intelligence, Machine Learning, Computer Vision, NLP, or a closely related field. (Bachelor’s degrees with an exceptional, proven research track in Multimodal AI will also be considered).
- Experience: 2+ years of hands-on experience in deep learning research or applied AI engineering.
- Track Record: Prior work or an active open-source portfolio in chart/document understanding, Visual Question Answering (VQA), fact verification, or grounded text generation.
- Academic Literacy: Proven experience running ablation studies, tracking machine learning experiments, and formatting/writing research papers to publication standards.
- Technical Skills & Frameworks
- 1. Core Machine Learning & VLMs
- Languages & Core Frameworks: Advanced Python, PyTorch (≥ 2.0), HuggingFace Transformers.
- Vision-Language Models (Mandatory): Hands-on experience with Qwen2-VL (2B/7B) and InternVL2 series.
- Advantageous Baselines: Experience with LLaVA, BLIP-2, MatCha, ChartLLaMA, or benchmarks like ChartQA, PlotQA, and DVQA.
- Fine-Tuning: Parameter-Efficient Fine-Tuning (PEFT / LoRA).
- 2. GPU Computing & Architecture
- Hardware Management: CUDA, multi-GPU setups (FSDP, DeepSpeed), and mixed-precision training (FP16 / BF16).
- Visual Grounding: Grid-based partitioning, region-level attention, and bounding-box supervision.
- Data Pipelines: Multimodal dataset management (image-text pairs), data augmentation, and error-injection pipelines.
- 3. Evaluation & Tracking
- Metrics: Lexical and semantic NLP metrics via sacrebleu, rouge-score, and bert-score.
- Frameworks: LLM-as-a-Judge frameworks and custom automated factual-consistency scoring.
- Experiment Tracking: Weights & Biases (W&B) or MLflow.
- 4. Academic Writing
- Documentation: LaTeX (Overleaf) utilizing template styles for IEEE, Springer, or ACL venues.
- Communication: Ability to write clear, concise technical reports detailing the outcomes of each experimental phase.
Contact Detail : 95661 33822
Job Type: Part-time
Pay: From ₹10,000.00 per month
Benefits:
Work Location: Remote
