app/data/portfolio_data.txt

Personal background
Ishmeet is a 21-year-old CSE student who has slowly shaped life around curiosity, discipline, and a steady love for building things that actually work. Growing up, there was always a pull toward logical problem-solving and patterns, whether in mathematics, puzzles, or the way software behaves when it is pushed, broken, and fixed again. Over time, that abstract curiosity turned into a focused interest in artificial intelligence and machine learning, especially in how these systems can read, structure, and reason over human language.
​

Her personality sits at an interesting intersection of ambition and reflection. There is a drive to push harder with every semester, every project, and every new concept, but also a strong tendency to pause and deeply understand why something works, not just how. This blend shows up in the way she studies, codes, and plans: she does not just collect tools but tries to build a consistent mental model of compilers, learning algorithms, language models, and software systems as a whole. That mindset naturally supports both academic performance and independent projects, forming the backbone of her journey into AI and software engineering.

Education and academics
Ishmeet is in the final year of a B.Tech in Computer Science and Engineering with a specialization in Artificial Intelligence and Machine Learning under the PTU 2020 syllabus. Semester by semester, she has built a strong academic record with SGPAs in the 7.3–8.7 band, including 7.86 in semester 4 (21/21 credits), 8.70 in semester 5 (23/23 credits), and 8.59 in semester 6 (22/22 credits). These numbers quietly reflect both consistency and growth: not just passing exams, but steadily improving understanding in core subjects and advanced topics.

Her coursework has spanned traditional CS foundations along with modern AI-focused subjects. Alongside data structures, algorithms, operating systems, and database systems, she has engaged deeply with machine learning, neural networks, and natural language processing. This means she does not treat AI as a black box; instead, she balances intuition with mathematical reasoning and tries to connect theory to implementation details in code. Over time, this has created a foundation that supports serious experimentation with models, from classical ML to transformers.

Interest in AI, ML, and NLP
From early in her degree, AI and machine learning captured her attention not only as trendy fields but as powerful, formal systems that can be understood, tuned, and deployed thoughtfully. She has sought out graduate-level, in-depth explanations of core ML concepts such as estimators, bias–variance trade-offs, maximum likelihood estimation, Bayesian statistics, supervised learning, unsupervised learning, and stochastic gradient descent. This means she does not just apply algorithms mechanically; she tries to understand what the model is actually doing in parameter space and how training dynamics respond to data quality, regularization, and optimization settings.

Within AI, natural language processing holds a special place. She is particularly drawn to text classification, emotion and sentiment analysis, retrieval-augmented generation, and the broader question of how machines can model and generate human language in contextually meaningful ways. This interest has driven her to build end-to-end systems that clean, encode, and interpret language data, and then expose those capabilities through usable applications. Over time, this has matured into a focus on LLM fine-tuning, optimization, and evaluation for real-world use cases.

Projects in NLP and deep learning
One of Ishmeet’s flagship projects is an end-to-end sentiment analysis web application that takes flight-related reviews from users and classifies them as positive, negative, or neutral in real time. She designed the pipeline to include text preprocessing steps such as tokenization, stopword removal, normalization, and transformation into model-ready features, followed by ML or NLP models that perform the actual sentiment classification. The front end exposes a clean, accessible interface, while the backend handles inference and returns clear visual feedback on sentiment predictions, deployed on Netlify for easy public access.
​

Another core project is a Harry Potter–themed retrieval-augmented generation chatbot hosted as a Hugging Face Space. This system allows fans to ask questions about the Harry Potter universe and receive grounded, context-aware answers instead of vague or hallucinated responses. Under the hood, she built and fine-tuned a T5/FLAN-based NLP model into a chatbot, integrated a RAG pipeline that retrieves relevant passages from a curated corpus, and structured responses so that they rely on retrieved context rather than just parametric memory. She also experimented with prompt templates, context window management, and model configuration to balance answer quality, latency, and hallucination control, turning the project into a real testbed for applied LLM techniques.
​

Internships and professional experience
Ishmeet’s professional journey so far is shaped by two core experiences that blend automation, AI, and real-world problem-solving: her Python automation internship at Roots Analysis Ltd in Mohali and her AI research role at Vyza Solutions. Together, these roles show how she moves from theory and classroom concepts into production-focused solutions that save time, improve accuracy, and directly support business and operational goals.
​

At Roots Analysis Ltd, she worked as a Data Analyst (Python Automation Intern) from January 2025 to September 2025, focusing on building AI-powered automation for large-scale data extraction and processing. Her primary responsibility was to design and implement intelligent web scrapers that could reliably collect structured information from complex, inconsistent websites at scale, significantly reducing the manual effort previously required for research and data gathering. Through this work, she built scalable data pipelines that not only accelerated data collection but also improved data accuracy and made it easier for multiple departments to access clean, ready-to-use information.
​

The impact of this internship can be seen in both quantified and qualitative improvements. The intelligent web scrapers she developed reduced repetitive manual work by as much as 80%, freeing analysts and researchers to focus on higher-level tasks instead of low-level data copying and cleaning. These automations helped accelerate research timelines, meaning projects could move from data acquisition to analysis much faster, ultimately increasing overall team efficiency and enabling better decision-making across the organization. Her contributions reinforced the idea that thoughtful automation can transform workflows, not just marginally speed them up.
​

This work also exposed her to real-world technical and operational challenges. She had to deal with inconsistent website layouts, frequent structural changes, rate limits, and the difficulty of processing large datasets without overwhelming systems or violating constraints. To handle these issues, she developed adaptive scraping logic that could adjust to layout changes, implemented task parallelization to speed up processing safely, integrated robust retry mechanisms to make the pipelines resilient, and added automated data validation to maintain data quality. These solutions required her to combine Python skills, problem-solving, and a solid understanding of both web technologies and data engineering principles.
​

Before and alongside this, Ishmeet worked with Vyza Solutions as an AI Researcher from September 2024 to January 2025, where she applied machine learning to the domain of hospital automation. Her main project involved building an AI model to predict patient inflow based on historical trends and real-world behavior, helping hospitals anticipate demand more accurately. She also engineered a resource allocation algorithm aimed at optimizing the use of medical staff and equipment, aligning resources with predicted patient loads so that care delivery could be smoother and more efficient.
​

The benefits of this work were significant in an operational context. By improving the accuracy of scheduling and resource planning, her models contributed to reducing patient waiting times and smoothing bottlenecks in hospital workflows. The system she helped design supported administrators and planners in making data-driven decisions about staffing, appointment slots, and equipment usage, translating statistical insights into tangible improvements in everyday hospital functioning. For an AI engineer in training, it was a powerful demonstration of how models can directly shape human experiences in critical environments like healthcare.
​

The challenges at Vyza Solutions were primarily related to the nature of medical data and the unpredictability of real-world patient inflow. She had to work with limited historical datasets, irregular patterns in patient arrivals, and noisy, imperfect data that did not always follow clean statistical assumptions. To address these obstacles, she performed careful feature engineering, applied smoothing and normalization techniques to stabilize signals, experimented with ensemble models for more robust predictions, and validated results using multiple forecasting approaches to ensure reliability. This process deepened her understanding of how to design, evaluate, and harden models intended for practical deployment rather than controlled academic settings.
​

Taken together, these internships illustrate Ishmeet’s growth from a student learning algorithms in theory to a practitioner capable of designing and deploying AI and automation systems in production-like environments. At Roots Analysis, she focused on building resilient automation pipelines that fundamentally reshaped how data flowed through the organization, while at Vyza Solutions, she engaged with predictive modeling and optimization in a high-stakes healthcare context. Both experiences highlight her ability to own end-to-end problem-solving: understanding requirements, designing solutions, handling real-world constraints, and delivering systems that measurably improve efficiency and outcomes

Broader technical skill set
Beyond individual projects, Ishmeet has developed a broad and practical technical skill set that supports real-world AI engineering. She writes most of her work in Python, using it for everything from data preprocessing scripts to training loops and API backends. She is comfortable with GPU-based workloads and has access to an NVIDIA GeForce RTX 2050 GPU, which she uses to train, fine-tune, and evaluate models locally when possible. This capability helps her understand the details of GPU utilization, batch sizing, and performance trade-offs in model training.

Her tooling includes VS Code, Jupyter Notebook, Git and GitHub, and Python virtual environments, all of which she uses regularly for development and experimentation. She also works with web technologies to deploy ML-powered applications, such as React-based front ends and Netlify deployment workflows, enabling end-to-end ownership from model to user-facing interface. This combination of backend ML knowledge and frontend deployment skills makes her well-suited for AI engineering roles that span data, models, and product.
​

Learning style and self-driven growth
Ishmeet’s learning style is characterized by deliberate practice, curiosity, and a willingness to go beyond the syllabus. She seldom settles for surface-level understanding and often seeks graduate-level explanations of topics, including rigorous treatments of estimators, bias–variance, and optimization theory. This pattern shows in the way she revisits lecture material, consults external references, and translates theory into experiments in code, testing variant models and hyperparameters to see how behavior changes in practice.

She also tends to interleave multiple topics: while working on NLP models, she might simultaneously be revisiting compiler design or exploring how graph-based optimizations resemble computational graphs in deep learning frameworks. This cross-pollination of ideas helps build a more unified mental model of computing and learning, where concepts from one area often clarify or strengthen understanding in another. Over time, this has cultivated both depth and flexibility in how she approaches new technologies or research directions.

Work–life balance and daily routine
Balancing late-night coding, academic responsibilities, and personal life is an ongoing theme in her journey. She often spends long stretches in front of a screen debugging, training models, or reading documentation, but she is increasingly intentional about maintaining a healthier rhythm. Her daily and weekly routines are evolving to include planned study blocks, defined coding sessions, and more structured breaks, rather than letting tasks spill endlessly into the night.

Health and fitness are important anchors for this balance. Ishmeet has explicit goals to build strength, look toned, and stay healthy, which means carving out time for physical activity even during intense academic or internship periods. These fitness goals help counterbalance the sedentary nature of coding and studying, reinforcing the idea that long-term success in tech requires sustainable energy and wellbeing. She is learning to respect her own limits and to see rest, movement, and offline time as part of a productive system rather than a distraction from it.
Hobbies, interests, and influences
Outside strictly academic or professional work, Ishmeet enjoys engaging with content that stimulates imagination and curiosity. She spends time on platforms like YouTube, Prime Video, and Hotstar, often leaning toward stories and genres that provide both escape and inspiration. The Harry Potter universe, in particular, has been a meaningful fictional world, which later translated into a concrete technical project through the Harry Potter RAG chatbot. This pattern—turning personal fascinations into structured technical challenges—shows how her hobbies and projects reinforce one another.
​

She also finds satisfaction in the process of coding and debugging itself, enjoying the problem-solving flow that comes with diagnosing issues, reading stack traces, and iteratively refining solutions. Late-night coding sessions are common, but they are not just about grinding; they also represent a quiet, focused environment where she can explore, learn, and build without interruption.This habit, paired with deliberate attempts at better balance, continues to shape how she grows as both a developer and a person.

Professional identity and future goals
As her degree nears completion, Ishmeet is clearly oriented toward a career as an AI engineer or ML-focused software engineer. She is actively preparing for placements and roles that will let her design, build, and deploy models that solve real problems. She wants to be involved in work that blends research-driven thinking with production-level impact, whether in NLP, document intelligence, analytics, or smarter interfaces powered by language models.

In the near term, she aims to keep expanding her portfolio with high-quality, well-documented projects that demonstrate both technical depth and end-to-end execution. That includes improving and showcasing the sentiment analysis app, the Harry Potter RAG chatbot, and the upcoming personal RAG-powered portfolio site. Longer term, she is interested in pushing deeper into advanced NLP, offline LLM deployment, and model generalization, along with continued growth in areas like computer vision, clinical text analysis, and research-grade experimentation. Step by step, she is building a trajectory that turns curiosity and discipline into a sustainable, impactful career in AI.