Activity
9K followers
-
Akhil Bidhuri shared thisbuildspace gave me this #NFT for completing the project using #solidity language on #ethereum #eth https://lnkd.in/gFRRW5wN
-
-
Akhil Bidhuri shared thisCleared REST API certification assessment from Hackerrank.
-
Akhil Bidhuri shared thisFrustrated of approaching passive candidates? Try Instahyre and get the best talent within seconds! Check out the link below.
-
Akhil Bidhuri liked thisAkhil Bidhuri liked thisOne of the nicest quality-of-life additions in Go is sync.WaitGroup.Go(). A small change that removes boilerplate, improves readability, and makes concurrent code a little harder to get wrong. It's always interesting to see how Go continues to improve developer experience while keeping things simple and explicit. #golang
-
Akhil Bidhuri liked thisAkhil Bidhuri liked thisPrecise, valid, to the point technical docs on how to run X no of GPU's as one. No BS, marketing. Pure tech. Love this from Factryze team. https://lnkd.in/g7bYHeTc #gpus #networking
-
Akhil Bidhuri liked thisAkhil Bidhuri liked thisC might become more used in upcoming years than C++, for developing modern numerical solvers. Specifically because once we write our solvers in C, we can easily run it on GPUs, using either OpenCL, or CUDA. If we write CUDA kernels in C++, it will only run on Nvidia hardware. Thus vendor locking forever.
-
Akhil Bidhuri liked thisAkhil Bidhuri liked thisOne of those engineering moments that reminded me how amazing modern databases are. We recently had to move ~17 million production product records in PostgreSQL. Our initial approach was pretty straightforward: - application-level inserts - all indexes kept active during migration It worked, but the migration time was looking like 4+ hours. Then we changed just two things: - switched to PostgreSQL COPY - created indexes after the data load instead of during inserts The result was honestly wild: - the actual data copy finished in ~10 minutes - even after rebuilding indexes, the total migration time was still more than 10x faster The interesting part was understanding why. With row-by-row inserts, PostgreSQL has to: - update indexes for every single row - handle transaction overhead repeatedly - do far more random disk IO - process a lot more network/application round trips At scale, all of this adds up very quickly. COPY works very differently. Instead of treating every row like an independent operation, PostgreSQL starts processing data as a large sequential stream. That means: - much fewer network round trips - larger internal batching - more sequential disk writes instead of scattered writes - better memory and WAL utilization - significantly lower per-row overhead And the index part was even more interesting. Maintaining indexes during millions of inserts means PostgreSQL keeps rebalancing and updating B-trees continuously for every row being written. Creating indexes after the bulk load lets PostgreSQL build them in a far more optimized sequential manner, which is dramatically cheaper. #database #postgresql
-
Akhil Bidhuri liked thisAkhil Bidhuri liked thisWith a cup of coffee in my hand, rains outside, I was just trying to perform a transaction (bundle of multiple operations) in my #aws #dynamodb table. Simple! Although my work was done, it led me to wonder what would have happened if someone else had also tried some other logic on the exact variable ID I was working on. Would that have made things inconsistent? Well, for sure we know, that this case of multiple transactions happening at the same time would already have been taken care of.. but what excited me to leave the cup of coffee was to understand how. How can concurrent transactions be handled with dynamodb?Versioning in case of DynamoDB concurrent transactionsVersioning in case of DynamoDB concurrent transactions
-
Akhil Bidhuri liked thisAkhil Bidhuri liked thisPramatra Space has been building quietly. When we decided to make photonics chips for quantum secure satellites - people called it a moonshot. We have it in our hands now ! From air defence, to earth observation, to designing an entanglement-based photonics chip for satellite (QKD) Quantum Key Distribution at Pramatra Space - has been the most adventurous and most certain choice of my career. This journey has been a roller coaster, and building resilience is the biggest reward. Today we have closed our pre-seed round, with the chip designed by us and tested at IIT Madras. My sincere thanks to our incoming investors from seafund , Rebalance , Magnivia Ventures / Peaceful Progress, and Awais Ahmed. And my heartfelt gratitude to all our early backers - Techstars Space Accelerator, Neel Mehta, Raghu Das, Bhavya Doshi and every early believer who backed us - family, friends, advisors, team and partners. Thank you for trusting us through every turn and trough. We're just beginning to accelerate further. Vinay Hukumchand Sheema Egbert | Steve Suarez® Mani Thiru Sarvagya Dwivedi Sharad Sanghi Abhishek Shinde Aishwarya Malhi Vikas Kumar Matt Kozlov Narendra Bhandari MANOJ KUMAR AGARWAL Mayuresh Raut Agam Gupta Sriram Sastrigal SunilKumar (Singhvi) https://lnkd.in/gHja7HkvPramatra Space Raises Pre-Seed Funding To Build a Quantum-Secure Future From Space - Pramatra Space | Quantum Key DistributionPramatra Space Raises Pre-Seed Funding To Build a Quantum-Secure Future From Space - Pramatra Space | Quantum Key Distribution
-
Akhil Bidhuri reacted on thisAkhil Bidhuri reacted on thisAfter an incredible journey at Careem for 7 years, I'm navigating a chapter change due to recent restructuring that happened. I'm deeply grateful to the leadership, my teams, and our partners who made it possible to build and scale platforms that made an impact across food, quick commerce, smart mobility, partner platforms and beyond. Over the last decade, I've built and scaled systems and teams at Careem, Amazon and Zomato from 0→1 and 1→N across both sides of the marketplace: customer experience, seller/merchant platforms, and everything in between. From food and quick commerce to smart mobility, IoT-connected fleets, and partner platforms - I've worked at the intersection of commerce and movement, where reliability, speed, and scale are non-negotiable. I'm now actively exploring my next engineering leadership role, whether that's doubling down on a domain I know deeply, or stepping into something new where strong platform thinking and team-building instincts travel well. If you're building at scale and need someone who can hold both the technical and the human side of the equation, let's talk. To anyone navigating a similar transition right now and needs support I'm opening up my topmate.io calendar(Link in comments) for the next 30 days, with one session each day dedicated to helping people find their footing and their next role. And any platform fees earned from these sessions will go straight to a cause worth supporting, I'll be donating them in full.
Experience & Education
-
Careem
View Akhil’s full experience
See their title, tenure and more.
or
Licenses & Certifications
Publications
-
Enterprise Chat Platform using Machine Learning Techniques
IJIRSET
See publicationEnterprise chat platform is a web application designed on the principles of machine learning. It allows
each on boarded enterprise to communicate with their employees by using group chats. The system can be accessed
from either a desktop or a mobile device. The proposed chat platform will include two additional features apart from
basic chat features like chatting, calling and file sharing namely sentiment analysis and text summarization. Text
summarization, a NLP feature, creates…Enterprise chat platform is a web application designed on the principles of machine learning. It allows
each on boarded enterprise to communicate with their employees by using group chats. The system can be accessed
from either a desktop or a mobile device. The proposed chat platform will include two additional features apart from
basic chat features like chatting, calling and file sharing namely sentiment analysis and text summarization. Text
summarization, a NLP feature, creates precise and short summaries for a given source document. The Encoder-Decoder
RNN architecture developed has proven to be effective when applied to a text summarization problem. Sentiment
Analysis intends to gage a person‟s opinion on a project or help determine the outlook of a speaker regarding some
contextual data. The CNN technique is an applicable algorithm to classify and interpret sentiment of data.
Projects
-
PingU
- Present
See projectEnterprise Chat Platform with Text summarisation and sentiment analysis. Dashboard for managers to manage and visualize their teams. With the real time capabilities of a chat platform.
-
Half cheetah - Reinforcement learning.
Applied ARS(augmented random search) technique described in the paper released in March 2018 by Horia Mania and others. To OpenAI gyms Half Cheetah environment and trained it to walk and obtained results within a training period of just 3 hours on a CPU.
-
Stock price prediction
A stock price predicting application which predicts the stock price and visualized the stock. Developed using Flask(Python web framework ),MySQL,Scikit learn, Chartist.js.
Regression techniques were applied and features such as past prices, GDP, Sensex, NIFTY, Govt. etc were considered. -
Virtual mouse
We all have seen sci-fi movies like Ironman and wondered when will this kind of technology become a reality. Taking inspiration from there I developed an application in Python using computer vision through which you don't need a mouse anymore to control your machine.
Control the machine using your hand gestures. Front camera captures the frames and the algorithm determines the gestures and based on gesture action to be performed is decided like right click, left click , scroll up and scroll…We all have seen sci-fi movies like Ironman and wondered when will this kind of technology become a reality. Taking inspiration from there I developed an application in Python using computer vision through which you don't need a mouse anymore to control your machine.
Control the machine using your hand gestures. Front camera captures the frames and the algorithm determines the gestures and based on gesture action to be performed is decided like right click, left click , scroll up and scroll down. -
Jawaab
-
Jawaab is an question answering tool which can answer questions on any given resource(url or file)
-
Picsta
-
See projectPicsta is a social networking web app built using MEVN stack. It has features of following other users, like, comment, post, search. Still work can be done to make it more real-time, scalable, feature rich and efficient.
Honors & Awards
-
Winner CourseHero 2023 hackathon
CourseHero
Won first place for question answering app that I made in the CourseHero 2023 hackathon.
-
Connected Recognition.
Cisco
CR - Innovate everywhere. For work on the Ganesha Chatbot.
-
Connected Recognition.
Cisco
CR - Innovate everywhere. For work on Lab runs and resource tool.
-
Top 6 in H_Acko(Acko Hackathon)
Acko
Were selected as one of the top 6 teams at h_acko. Built a PWA ( Progressive Web App), which provides short term insurance to the users and price based on risk analysis of the customer. The system also had features like the referral system which indirectly helps in marking and reduces the next premium payment for the customer by a small margin. the App primarily focused on getting more customers upon introduction.
-
National level winner in Hack2innovate by T-Hub and NITI Ayog
T-Hub ,NITI Ayog and PWC
Won at the national level Hack2innovate for the problem weather prediction by PWC .
Hack2innovate organized by T-Hub and NITI Ayog.
-
First place in Cryptothon by Unocoin
Unocoin
Winner of first place at Cryptothon conducted by Unocoin.
Built an android game similar to pokemon go from scratch in 24 hours to promote bitcoins in India(re imagining the future of bitcoins).
Other similar profiles
Explore more posts
-
Kamran Ali
Arpit sitting in Nagpur waits for 8 seconds for an App to load video Frustrated he deleted the App and gave 1 star review. How to fix it? Solution: Content Delivery Networks (CDNs) What is a CDN? 1. Distributed network of servers placed strategically around the globe 2. Stores cached copies of your content (images, videos, static files) closer to users 3. Routes user requests to the nearest server instead of your origin server How CDNs Work 1. User requests content from application 2. CDN intercepts the request and checks its nearest edge server 3. If content exists in cache, it's served immediately 4. If not cached, CDN fetches from origin server and caches for future requests 5. Subsequent users in that region get lightning-fast responses Key Benefits: 1. Reduced Latency 2. Improved Performance 3. Better Scalability 4. Enhanced Reliability via multiple edge locations 🤔 over to you, what are downsides of CDN? Let me know in the comments! 👇 --- You've got System Design skills but showcasing them in 60 minutes interview is a completely different challenge for e.g. where to start? How much time to spend on HLD vs LLD etc Let me share a FREE 7-day email course that provides a step-by-step framework for acing your 60-minute System Design interview 📣 Sign up for the course at systemdesign.engineer, to learn time allocation to demonstrate full range of design skills
22 Comments -
Cortex
If you give everyone the same size shoe, everyone walks funny. 👟 Pragya Jaiswal, Engineering Manager at Paxos, shared at #IDPCON2025 why "one-size-fits-all" is the fastest way to kill your IDP adoption. To win over developers, Paxos pivoted to: 👥 Persona-based workflows (SREs ≠ Product Engineers). 📔 Actionable scorecards (No "walls of shame"). 🔥 High-impact use cases like scaffolding. The result? 100% awareness across the org. See how they did it: https://lnkd.in/g_me9PkR
-
Mehul Sachdeva
When millions tune in to a cricket final on Hotstar or a new season drops on Netflix, the servers don’t wait for the rush, they’re already scaled up. We learned this the hard way. Once, during a flash sale, we relied on “smart” autoscaling. The plan was simple, when CPU spikes, just add more servers. Sounds solid, right? Except when traffic hit, latency exploded. Why? The new servers needed over a minute to boot and warm caches. Meanwhile, the old ones, already choking had to juggle both real users and the heavy lifting of onboarding newcomers. Scaling up only made things worse. The real trick? Scale before the storm. Keep some warm standbys, bring new servers in gradually, and watch queue depth, not just CPU. #technology #netflix #hotstar #programming
-
Business Review Live
Runpod hits $120 Mn annual revenue run rate as developer-focused AI hosting scales globally “It was almost two years where we really didn’t have any funding,” Zhen Lu, founder, Runpod said. Consequently, Runpod never introduced a free tier. The platform had to cover its own costs, even if profits remained modest. Unlike other AI cloud services that originated from crypto mining, the founders deliberately avoided taking on debt. Read more :- https://lnkd.in/gzmVusUa #AIInfrastructure #ArtificialIntelligence #CloudComputing #DeveloperTools #TechGrowth
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore More