AI Engineer with less than a year in Computer Vision & NLP
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
Harshit Sojitra is an aspiring AI Engineer with a strong academic background, currently pursuing an M.Tech in Computer Science & Engineering. With practical experience as a Web Development Intern and hands-on exposure to AI/ML, Deep Learning, and Generative AI, Harshit demonstrates proficiency in Python, C++, and various frameworks like PyTorch and TensorFlow. Their project portfolio highlights work in semantic communication with Generative AI, underwater image enhancement, and LLM-based ranking systems, showcasing a keen interest in developing innovative AI solutions.
Indian Institute of Information Technology, Vadodara
M.Tech · Computer Science & Engineering
August 1, 2025 – August 1, 2025
Shantilal Shah Engineering College, Bhavnagar
B.Tech · Information Technology
July 1, 2021 – May 1, 2025
Shri GyanBharti School, Junagadh
12th
June 1, 2020 – May 1, 2021
Shree Swaminarayan Gurukul, Visavadar
10th
July 1, 2018 – May 1, 2019
Virtual Rudra
Web Development Intern
January 1, 2025 – May 1, 2025
India
Multi-Scale Transformer-Based Underwater Image Enhancement
June 1, 2026 – Present
• Designed a CNN encoder-decoder multi-scale transformer network with cross-attention for enhancing degraded underwater images. • Applied feature fusion (FSM) and residual learning to improve color correction and image clarity.
Generative AI for Text-to-Image Generation with Trajectory Control
June 1, 2026 – Present
• Investigated training-free layout control in Stable Diffusion, evaluating trajectory-conditioned image generation methodologies against traditional mask and box-based baselines. • Optimized spatial conditioning in pre-trained generative models by manipulating text-related cross-attention layers without requiring additional specialized training pipelines.
LLM-based Full Ranking System for Passage Retrieval
June 1, 2026 – Present
• Built a long-context LLM-based passage ranking system, replacing sliding-window methods with full-ranking to improve efficiency and ranking accuracy. • In depicted importance-aware loss and multi-pass label generation, we achieve better NDCG performance and reduced latency.
Semantic Communication with Generative AI
June 1, 2026 – Present
• An Image Caption Neural Network translates raw visual data into text, allowing a Large Language Model to assign contextual importance scores that follow adaptive, error-resilient transmission over noisy channels. • BERT mathematically predicts and repairs any corrupted text, which a Generative AI model then utilizes to synthesize a high-quality reconstruction of the original image.
Fundamentals of Deep Learning
NVIDIA
October 1, 2025 – Present
Cultural Fit Analysis
The candidate's project portfolio is highly focused on advanced AI/ML research, demonstrating a strong passion and deep interest in the field. This aligns well with a culture that values innovation and technical depth in AI. The academic background and project diversity suggest a proactive learning attitude. However, the lack of significant industry experience beyond a web development internship means there's limited data to assess adaptability to corporate environments or cross-functional team collaboration.
Soft Skills & Operational Fit
The candidate's project descriptions indicate an ability to work on complex, research-oriented problems, suggesting strong problem-solving and analytical skills. The academic nature of most projects implies a capacity for independent research and learning. However, there is limited evidence of collaborative work or communication skills in a professional team setting, which are crucial for operational fit.