Headstarter

[comment your email if you are applying for ai]...Software engineers should know how hiring managers think when hiring in AI. A recent YC founder told me: "Someone should be able to come in and like build a production llm app and defend their choices in terms of model, rag pipeline simple and advanced, retrieval strategy, etc" Two keywords here is "defend" and "production" **Defend** stems from confidence, conviction and opinion that "this" was is right this being the llm model or agent workflow setup. **Production** means the code is customer-first, business logic driven and scalable beyond just prototype and POC. Production insight matters a lot because at scale there usually exists some linear fixed cost and limited real-time support ie no more than 75-100 openai calls or llama-3 calls via Openrouter, Anyscale at one given time I am fortunate enough that many of my friends today have been hiring after YC or being venture backed and usually getting an ai intern is their first bet. IF you think this is an area of interest for you, I would love to make few referrals. Comment "ME" and your email!

Introducing Headstarter Project 5: AI Rate My Professor - Proffinder Crerated by Team #M³: Mathew Xie, Matthew Floresca Tom and me Using Next.JS, React.JS, Material UI, Node.JS, OpenAI API, Pinecone, and Vercel 🔗 Proffinder App: https://lnkd.in/g3A6ivPS 📊 Team Presentation: https://lnkd.in/emUHyZcf 🎥 Full Demo: https://lnkd.in/ewvM3Rfb 📝 Reddit Discussion: https://lnkd.in/e_3_Z67V 💻 Code: https://lnkd.in/egevhmQC Proffinder allows students to search for professors through a chatbot. Users can filter professors based on concentration and review stars. Our AI system gives our users the best professors, matching what our users want. If our answers are unsatisfactory, users may suggest new professors our AI can review. Proffinder has a vast knowledge bank of professors from whom to make suggestions. Therefore, users will always get the best recommendation for what they’re looking for. If an answer is inadequate, users may provide the ID of a professor they prefer to proffinder. Proffinder will scrape information about the professor and add the reviews to the database. Proffinder works 24/7 and provides quick answers to user inputs. The maximum wait time for a response is ~5 seconds. A huge thank you to Yasin Ehsan 🚀 and Faizan Ahmed for this amazing project. #Headstarter #SofftwareEngineering #ComputerScience #HeadStarterFellow #SWE #Fellowship #Internship #Learning #LearningJourney #Tech #Projects #AI #GenerativeAI #LinkedIn OpenAI Vercel Pinecone

Hey everyone! Week 5 of Headstarter's SWE Fellowship was a big one. It was the last week of regular instruction from our Project Lead, Bill. Now, it's time to focus on our final projects. 𝗛𝗲𝗿𝗲'𝘀 𝗮 𝗹𝗼𝗼𝗸 𝗯𝗮𝗰𝗸 𝗮𝘁 𝘄𝗵𝗮𝘁 𝗵𝗮𝗽𝗽𝗲𝗻𝗲𝗱 𝗶𝗻 𝗪𝗲𝗲𝗸 𝟱: On 𝗠𝗼𝗻𝗱𝗮𝘆, Faizan held a Q&A session where he discussed what startups look for in a hire (including red flags) and where to apply. He also told us about Pitch and Run, and how groups like it can provide effective opportunities for networking. On 𝗧𝘂𝗲𝘀𝗱𝗮𝘆, Jack O'Brien visited with fellows for the final time. He advised to identify your niche and make the service your app provides hyper-relevant to your target audience: “𝘍𝘰𝘤𝘶𝘴 𝘺𝘰𝘶𝘳 𝘶𝘴𝘦𝘳 𝘰𝘯 𝘴𝘰𝘮𝘦𝘰𝘯𝘦 𝘳𝘦𝘢𝘭𝘭𝘺 𝘴𝘱𝘦𝘤𝘪𝘧𝘪𝘤. 𝘐𝘧 𝘺𝘰𝘶 𝘴𝘢𝘺 𝘵𝘩𝘪𝘴 𝘪𝘴 𝘵𝘩𝘦 𝘢𝘱𝘱 𝘧𝘰𝘳 𝘦𝘷𝘦𝘳𝘺𝘰𝘯𝘦, 𝘪𝘵 𝘦𝘯𝘥𝘴 𝘶𝘱 𝘣𝘦𝘪𝘯𝘨 𝘢𝘯 𝘢𝘱𝘱 𝘧𝘰𝘳 𝘯𝘰 𝘰𝘯𝘦.” Afterwards, Faizan provided tips on how to improve UI/UX through A/B testing. He used Jungle, Magical Toys, Rilla, and PostHog as examples -- whose sites target specific users and use language and terminology that resonates with them. On 𝗪𝗲𝗱𝗻𝗲𝘀𝗱𝗮𝘆, Nael Ould Belkacem returned to show Track C fellows an example of an accepted pull request, and walked through the steps the contributor took when making it. On 𝗧𝗵𝘂𝗿𝘀𝗱𝗮𝘆, Headstarter held its weekly #Pathways2Tech, featuring: ▪️ Tony Tan (ex-Meta Staff SWE, Co-Founder and CEO @ Spaition, Inc.) ▪️ Abhi Vaidyanatha (Founding Product Marketer @ Convex) ▪️ Farhan Hossain (Founder and Technical Advisor @ noble. Crafters) On 𝗙𝗿𝗶𝗱𝗮𝘆, Headstarter kicked off its 2nd #HiringHackathon in two weeks (and third hackathon overall), featuring Rilla, Jelly, Pam, and Datalogz. Tariq Ahmed stopped by to offer his thoughts on what it takes to work well in teams and what makes a good hackathon project. On 𝗦𝗮𝘁𝘂𝗿𝗱𝗮𝘆, Faizan held a hackathon Q&A session where he discussed the do's and don'ts, the proper demo structure, and the importance of nice, neat UIs. We were then joined by iqram magdon-ismail (Co-Founder of Venmo, Co-Founder @ Jelly). He told us the story of how he came up with the idea for Venmo, invited Yasin to create a jelly with him, and gave advice on how to attain 1,000 users on your app. Afterwards, Abdullah Baig came by to tell us more about Pam, where they got the idea for the startup, and what they're looking for in a potential hire. He also shared the awesome store of how they came up with the name and the number one thing that made an engineer shine above the rest during his time at Amazon, Lyft, and Adobe. On 𝗦𝘂𝗻𝗱𝗮𝘆, fellows submitted their hackathon projects. Once the top 3 from each hiring track were selected, everyone hopped on a call where we were treated to 2 hours of hackathon demos, hosted by Faizan. 𝗧𝗟;𝗗𝗥: Week 5 out of 7: Complete. Check out some of the highlights below. 🚀 #ComputerScience #SWEFellowship2024 #HeadstarterAI

Want to get hired as a CS student? Show that you care. Here are 3 ways you can show that you CARE: - In your speech - In your resume - in your questions At any interaction you have with a hiring manager or engineer who you can glean insights from, do not appear aloof. Research them 30-sec before greeting them. Size a few ways to connect from mutual friends, alma mater, hobbies etc. Be present. Show all your attention when speaking and listening. Show that you CARE in your speech. So many people are quick to just apply. Job application. Peep. Pop. Applied and satisfied. As if it was a dopamine hit. Than add another data point to a minimal, pastel sankey diagram. In your resume, clear outline what meaningful contributions you did that can show ... product mindset, customer experience and business requirements. This means having metrics, outcome based remarks than "I know java, c++, etc" In your resume, word it for every archetype of reader: the recruiter, the engineer, the hiring manager, etc word it for every attention-span of the reader: 6 seconds, 30 seconds, 2-min etc Show that you are well-read in resumes and have sieved the best points from a stash of "role-model" resumes Eliminate any possibility that you are oblivious, ignorant or "CARELESS." Show empathy. Show that you CARE in your resume. Trust me. You WILL get hired if you care. When asking questions, most people just do a russian-roulette. Or wing it. The response from the question often ends in giggles or nothing significant to peg a later conversation on. Like sea foam, the questions that are not thoughtful leave silently the same was they came. SO how do you ask GOOD questions? Simple. You just try. You CARE. Like a muscle the more you TRY the better you will get at it. Someone once told me "Quality of questions determines quality of life." Such a profound statement. I wholeheartedly support it. When you meet important people - in the context of career -, the more higher up they are the more they are likely to forget your face. BUT they will likely not forget your QUESTION especially if you made them FEEL a certain way. There is research that shows people who ASK questions at end of interviews are significantly more likely to get hired than those who are not. So show that you CARE by asking ***quality*** questions. So to recap, show that you care by: - In your speech - In your resume - in your questions Comment below what stood out to you about this post. Repost if you think others can benefit. Appreciate you for reading. 🚀 🚀 🚀

Most people want to be the BEST at everything. Successful people are ones who want to be exceptional at ONE thing.

Over the weekend I competed in my third Headstarter hackathon ! My team and I were tasked to create an app that places captions on videos for a company called Jelly Jelly (backed by the founder of Venmo😳). Can you believe we were selected as finalists in this hackathon as well?! This fellowship has really pushed me in ways I didn’t know were possible. It’s shown me how competitive I am, and how capable you can become with just a little pressure. Having essentially won my third hackathon, I’m definitely going to take a break from competing and hone in on the projects that my team and I have created plus ones I have as well. We actually got to talk with the founder of Jelly Jelly yesterday and tell him about our project and implemented an additional feature in our project to show him based on that call! As far as the future goes, I have been accepted into an externship with Webacy where I’ll be diving into blockchain development and security. I will also be continuing my studies in school, continue to make progress on the projects you guys have sort of been watching me build, and try to better document more things on here and YouTube ! Thanks for reading 🙏🏽 ( and if you happen to be looking to hire a dev, send me a message, or send this to someone that is!😀 )

Today I wrapped up 5 weeks as a Software Engineering Fellow at Headstarter. Throughout this period, I experienced tremendous growth in my technical skills, driven by intense learning and hands-on application of my skills in weekly projects. At the start of the second week, I felt lost and overwhelmed due to my unfamiliarity with the technologies used in the example projects. However, I quickly adapted by following along with the materials, reading documentation, and browsing online to fix bugs and add new features. This allowed my creativity to flourish as I worked on new projects. I would love to share some insights about the projects I’ve been working on and the key takeaways from the past few weeks! Here I go: 𝗜. 𝗣𝗮𝗻𝗠𝗮𝗻: https://panman.vercel.app/ PanMan helps you track pantry items, stay organized, and reduce food waste. With AI-generated recipes based on your pantry inventory, it supports all your cooking adventures. Video: https://lnkd.in/g2nfgD9f 👩🏻💻 Stack & Tools: Next.js, React, Material UI, Firebase, Gemini API, Vertex AI, Figma, Vercel. 🌟 Key Takeaway: If you don’t love the project you built, you’ll probably enjoy creating a better version. Initially, I built a bare minimum version and wasn’t satisfied with it. Later, I redesigned the UI, learned new tools, and built the version I had envisioned. 𝗜𝗜. 𝗤𝘂𝗶𝘇𝗪𝗶𝘇: https://lnkd.in/gvqKuzze QuizWiz uses AI to create flashcards from text prompts, making study sessions efficient and personalized. Video: https://lnkd.in/geKyGFuc 👩🏻💻 Stack & Tools: Next.js, React, Material UI, Firebase, OpenAI API, Clerk, Vercel. 🎉 Fun Fact: I might have binge-watched and read a bit too much Harry Potter before creating this, so don’t be surprised if some UI elements look like they belong in the wizarding world! 🧙♂️✨ 𝗜𝗜. 𝗦𝗼𝗺𝗺: https://lnkd.in/gsFdDn4b AI sommelier, powered by RAG (Retrieval-Augmented Generation), takes the guesswork out of wine selection by offering personalized recommendations based on your unique taste. Video: https://lnkd.in/gX_JDBdd 👩🏻💻 Stack & Tools: Python, Pandas, Pinecone, Jupyter Notebook, Kaggle, Next.js, React, Material UI, OpenAI API, Clerk, Vercel. 🌟 Key Takeaway: It’s perfectly fine—and often more exciting—to base your example projects on something you’re passionate about or interested in. ✨ Working on these projects has made me fall in love with the process of creating and building. Big thanks to Yasin Ehsan 🚀 and the entire Headstarter team for constantly pushing us to grow and for creating such a supportive developer community. I’m always eager to improve, so any feedback on my projects would be greatly appreciated🙏.

𝐌𝐜𝐆𝐢𝐥𝐥𝐢𝐚𝐧𝐬, 𝐥𝐢𝐬𝐭𝐞𝐧 𝐮𝐩! This one’s for YOU 🫵 Ever felt that sinking feeling of not knowing which course to take or which prof to choose? The confusion, the stress, the anxiety? 𝐅𝐞𝐚𝐫 𝐧𝐨 𝐦𝐨𝐫𝐞. This week at Headstarter, I made a website featuring an AI assistant, as a great add-on to #RateMyProfessors, designed to help you find the perfect prof for your needs. This project leverages RAG (Retrieval-Augmented Generation) to provide more accurate and relevant recommendations. Shoutout to Bill Zhang for his guidance 💻, Esterlin Jerez Paulino for his invaluable help 👓, and Erik Cupsa—his 𝘔𝘤𝘎𝘪𝘭𝘭 𝘌𝘹𝘢𝘮 𝘚𝘤𝘩𝘦𝘥𝘶𝘭𝘦𝘳 project was a huge source of inspiration ✏️. 𝐅𝐢𝐧𝐝 𝐲𝐨𝐮𝐫 𝐢𝐝𝐞𝐚𝐥 𝐩𝐫𝐨𝐟 𝐫𝐢𝐠𝐡𝐭 𝐡𝐞𝐫𝐞, 𝐫𝐢𝐠𝐡𝐭 𝐧𝐨𝐰 👉 https://lnkd.in/e47iG6iZ 𝘕𝘰𝘵𝘦: This is currently only an MVP. Due to OpenAI and Pinecone limitations, I couldn’t integrate all 2000+ datasets. If you’ve got solutions or want to contribute, you’re more than welcome! 𝐓𝐞𝐜𝐡 𝐬𝐭𝐚𝐜𝐤 𝐮𝐬𝐞𝐝: - Next.js - JavaScript & Python - Pinecone - Material UI - OpenAI API 𝐆𝐢𝐭𝐇𝐮𝐛 𝐫𝐞𝐩𝐨: https://lnkd.in/eX9vxqFA 𝐌𝐀𝐃𝐄 by McGill University: https://lnkd.in/eBhtrv3F #AI #McGill #StudentLife #FindMyProf

🌟 Introducing Sam, Your Helping Hands Chatbot! 🌟 I am thrilled to unveil Sam, this virtual assistant powered by cutting-edge AI technology! Developed using Google’s Gemini API, Sam intelligently responds to your queries as you navigate the Helping Hands platform—whether it’s making donations, finding information, or understanding our services. The AI responses are generated via Google AI Studio, ensuring that your experience is seamless and informative. 💬✨ Designed with responsiveness in mind, Sam offers a smooth experience across all devices, so you can get the help you need anytime, anywhere. This project was part of my fellowship at Headstarter, where I had the opportunity to build and integrate this chatbot using advanced AI tools. Proud to be part of a mission-driven initiative that makes giving easier and more impactful! 🌍💙 #HelpingHands #AIChatbot #GoogleGemini #HeadstartersAI #AIInAction #MakeADifference #ResponsiveDesign

I am sharing my progress in Weeks 3 and 4 at Headstarter. I learned and explored lots of different concepts and technologies. Came across many obstacles and bugs but all worth it. Had lots of fun making these.😋 Week 3 Project is a Basic Chatbot that I developed using llama3-8b and Vercel-AI SDK. I used Groq to implement this model in my project with the help of server side component provided by Nextjs. Styled it with Shadcn components.🌟 Week4 Project is a SAAS platform that provides AI generated flashcards to the user on any topic and lets them store it as well. Used Firebase Auth and Firestore. With Clerk for seamless SignIn and Authentication experience. And Stripe for payments and subscription. Styled it with Shadcn components.(I'll deploy it soon)✨