Founder × Intern
A high-fidelity chronicle documenting my 2-week AI & Prompt Engineering residency at Virtual Height. Mapping the interface between non-deterministic generative intelligence and deterministic system integration.
Featured Logs
Selected HighlightsInternship Timeline
Days 01 - 14Onboarding & Exploration: Mapping the Core AI & Python Blueprint
Onboarding at Virtual Height, exploring Python-based AI development pipelines, setting up sandboxes, and defining our core AI & Prompt Engineering curriculum.
Day 01: Foundations of AI, Python Lab, and the Velocity of Domain Mastery
Onboarding under Senior AI Trainer Rashmi, mapping the formal pillars of AI from ML to DL, mastering cross-platform Python execution, and demonstrating custom LLM and reverse engineering portfolios to the cohort.
Day 02: Python Control Flow, Functional Lab, and the Birthday Aura Hack
Setting up Python and VS Code envs, implementing conditional systems and match-case control flows, scripting a modular calculator, and presenting a surprise birthday web build to Rashmi.
Day 03: String Typology, Rule-Based Chatbots, and the Whiteboard Lecture
Analyzing list structures and dictionary lookups for chatbots, mastering string operations, and being selected by Rashmi to lecture the cohort on the whiteboard.
Day 04: APIs, JSON Schemas, and the Gemini AI Integration
Interfacing FastAPI with the Google Gemini API to build an asynchronous, type-safe chatbot, and mastering request-response structures and JSON payloads.
Day 05: Prompt Engineering Mechanics, Hyperparameters, and the AGI Paradigm
Deconstructing LLM prompting structures (RTACF), tokenization mechanisms, softmax temperature control, and exploring training, fine-tuning, embeddings, and RAG pipelines.
Weekend 01 Reflection: The Corporate Loop & The Majority Wins
Reflecting on the first five days of the internship. Realizing that while the technical teachings were profoundly basic, the true lesson was observing the endlessly 'cooked' corporate loop and the psychology of environments.
Day 06: HTML & CSS for Chatbot UI and the SagarAI Reaction
Transitioning to frontend interfaces by mastering HTML structures, CSS styling for chat bubbles, responsive design, and demonstrating the SagarAI chatbot.
Day 07: Flask Fundamentals & The Engineering Gap
Building backend bridges with Flask, handling HTTP requests, rendering HTML templates, and realizing the massive experience gap between traditional learning and production engineering.
Day 08: The AI Chatbot Project & Architectural Pragmatism
Integrating remote LLM Inference APIs with Flask to build a custom StudyBot, analyzing pipeline overhead, and handling execution exceptions with robust error handlers.
Day 09: Project Enhancement, State Management, and the Single-Page Chat Interface
Implementing session-based concurrent chat history storage in Flask, building a modular multi-personality system prompt compiler, and upgrading the user experience with asynchronous API requests.
Ecosystem Takeaways
Key Learnings
- Implementing concurrency-safe session state management to isolate user conversation histories.
- Designing request-response data flows between front-end inputs, back-end servers, and remote AI APIs.
- Instantiating local Flask server environments and declaring URL routes.
- Structuring webpage elements using HTML (forms, inputs, buttons).
- Recognizing the deep exhaustion and repetitiveness of the standard corporate 9-to-5 loop.
- Constructing structured prompt interfaces using the RTACF framework.
- Explaining and diagramming client-server API request/response cycles.
- Utilizing dictionaries for constant-time lookup maps in conversational routing.
Tools & Stack
- Flask Sessions
- Flask
- HTML
- Sociological Observation
- Gemini API
- FastAPI
- Python 3.11
- Fetch API
Challenges Overcome
- Securing session cookies and ensuring they modify state properly on nested object updates.
- Managing API connectivity limits and handling request connection timeout exceptions gracefully.
- Debugging Method Not Allowed (405) errors when forgetting to explicitly allow POST methods on routes.
- Aligning and floating chat bubbles appropriately for left/right dialogue flows.
- Maintaining focus and motivation when the curriculum is taught at a highly basic, introductory level.
- Optimizing prompt layouts to prevent context window token overflow.
Startup Integration Reflections
In a modern agency ecosystem, AI integration, Python automation, and Prompt Engineering are not peripheral utilities — they form the high-leverage engineering stack that bridges business logic with scalable generative automation. Aligning non-deterministic models using structured prompts and solid Python glue code is the absolute prerequisite for production-grade intelligence.