← Back to leaderboard
techwithakansha

Akansha Khandelwal

Verifiedactive
@techwithakansha

Software Engineer @Nagarro |50 k+ @LinkedIn |Tech and AI Content Creator | Sharing insights on AI ,Tech Tools ,No Code, Productivity| DM for collaboration

RosterAI Agents & Productivity ToolsAI Model News & Releases
14,209 followers · 466 following · Last polled 2h ago
Perf. Score
57.2
Rank #174
Value Score
90.6
#19 of 268 priced
Median views
8.6K
p25 5.8K · trailing 7d
Followers
14K
+0.0% 7d
ER (impr.)
0.85%
eng ÷ impressions
ER (foll.)
0.89%
eng ÷ followers
Est. CPM
$1.74
QT-rate basis
WoW views
+33.8%
▲ rising

Rates & value · est. from 8.6K median organic views

underpriced
Quote tweetbasis
$15
$1.74 / 1K views
Post
—
no CPM
Thread
—
no CPM
Retweet
$17
no CPM
Value Score 90.6 · rank #19 of 268 priced$200 / 1K engagementsrate age unknown (pre-audit import)

Trends

Follower growth
Median views / day
Engagement rate
View distribution · trailing-7d posts

Each blue dot is one organic post. The Performance Score ranks by the median (8.6K), not the mean (15K) — so a single viral post doesn't inflate a creator's reach. The p25 floor (5.8K) is the level 75% of posts clear.

Recent posts · ranked by views

PostPostedViewsLikesRepostsRepliesQuotesBookmarksER
I’ve been seeing coding models move beyond simply generating code, and IQuest-Q1 caught my attention. It’s a large, code-focused model built for developers, with a focus on stable performance across multi-scenario task evaluations. IQuest is evaluating it across four practical scenarios: → Building complex interactive applications through a game demo → 3D scene + spatial generation through a park-navigation style demo → Code diagnosis + repair → Complex work-task execution involving multimodal information + tools The final demo videos for all four scenarios will be available at release. I’m curious to see how IQuest-Q1 performs across these different workflows and what the demos reveal about the model’s capabilities. #coding #LLMfrozen6d ago48K17025130610.56%
The interesting part isn’t just connecting HeyReach to Claude. It’s being able to open Claude, describe who you want to reach + what you want to say, and have HeyReach set up and launch the LinkedIn outreach from your own accounts. No jumping between tools. No manually building every campaign. Just describe the campaign in Claude, let HeyReach handle the outreach, and focus on the replies. That’s a much more practical use of MCP.
frozen
1mo ago
42K
63
10
25
0
8
0.25%
AI voices can sound incredibly human, until they confidently pronounce one important name completely wrong. That’s what caught my attention about @OnepinAI. It checks every line of a voiceover for pronunciation, clarity and accuracy, then fixes the words it gets wrong directly in the same voice. No manually checking hundreds of lines. No re-rendering the entire take just because one word was wrong. A pretty useful QA layer for anyone shipping AI voiceovers.frozen6d ago39K1775411070.64%
Maybe the next evolution of software development isn’t just better coding tools, but better environments to build in. Base Code takes your repo, sets up the required cloud infrastructure, and gives the team a shared workspace. With humans and AI agents working in the same environment, the whole development workflow can look very different.frozen10d ago34K12341150130.56%
Claude can do the thinking. But turning that work into a polished, on-brand PowerPoint is where things usually get messy. @Templafy MCP changes that. Research + build your storyline in Claude, then let Templafy turn it into a real, editable PowerPoint using your company’s templates, brand rules + approved content. No copy-paste. No reformatting. Just a business-ready deck.frozen21d ago33K411213070.22%
AI builders should have the freedom to choose and swap the pieces of their stack as they build. The Nebius AI Builder Program brings the pieces together, from open models and Token Factory to runnable blueprints, working code, partner tools, and office hours with engineers. You also get $400+ in credits and discounts from day one, and it’s free to join. If you’re building your next AI project, this gives you a strong place to start. Join the Nebius AI Builder Program: https://t.co/eRzATa9aBZ #nebiuspartnershipfrozen28d ago33K10724230140.51%
What if your AI assistant could actually join the group chat? That’s what caught my attention about Rene. It can talk to your friends’ Renes, figure out a time that works for everyone, settle the usual group chat debates, and help keep plans moving, all inside iMessage. It also remembers you + adapts to your style, while taking care of things like shopping, emails, slides, and images. No app to download. Just text Rene like you’d text a friend.frozen21d ago33K646180100.30%
Reliable compute is quickly becoming one of the most important layers of the AI stack. GMI Cloud’s $668M Series B, led by ARCHIV with participation from NVIDIA, reflects how fast demand for GPU capacity and inference is growing as AI moves into real-world use. Contracted ARR is now more than 9x its year-end 2025 level, with approximately 4 trillion tokens processed every week. That’s meaningful scale. Excited to see what @gmi_cloud and @alex_yehya build next.frozen8d ago22K5758060.34%
Turn your market insights into ads that are ready to run. 🚀 Topview AI Marketer brings research, winning angles, creative concepts, and finished ad assets into one AI-powered workflow. Market research → Winning angles → Conversion-ready ads @TopviewAIhq https://t.co/OM0zyMLjTpfrozen29d ago21K2655050.19%
The interesting part isn’t just having multiple AI agents. It’s having a system that can coordinate them, learn from the process, and improve its own workflows over time. Raven takes agent orchestration to another level. 🔥frozen9d ago17K712040.08%
AI agents can write code. But can they actually understand what they’re looking at? PPTBench tests that missing piece: visual understanding → structured reconstruction → inspection → correction. The fact that only 2.57% passed every gate with zero recorded defects shows how hard Visual Coding still is.frozen9d ago16K551116070.55%
A simple brief can now become a complete video with @vizard_ai. What I like here is that creators can focus on the idea while the agent handles the planning, editing, and iterations. 6 tracks, $3,000 in cash + 100,000 points, and free to enter. Curious to see what creators come up with.frozen19d ago12K47119030.59%
AI-generated worlds are starting to feel less like videos and more like environments. LingBot-World 2.0 can respond to actions in real time, and its 1.3B Small Model can run on a single consumer GPU. That makes interactive world models much easier to experiment with and build on.frozen29d ago12K4710140100.69%
Joey vs Elon could be a one-off joke. Meme Arena makes it part of an endless interactive AI livestream, where viewers can bid to choose a fighter and the winner stays for another challenger. Go put your pick into the arena. The model behind it is Vivix W1. Quality nearly on par with SOTA models is appealing, but radically lower costs are what I'd look at first for a show that keeps generating. Blazing fast generation helps too. Expensive video gets difficult to justify once you want more than that first joke.2d ago12K37108040.50%
The coolest part isn’t just how he looks—it’s that he can respond and feel like a real character.frozen13d ago12K33712070.51%
The idea of keeping autoregressive quality while generating multiple tokens in parallel is seriously interesting. Uno takes a different approach to speeding up LLM inference: keep the original AR weights for quality, then add a lightweight diffusion adapter for faster generation. Up to 2.2× speedup with no quality loss makes this a very compelling direction for faster LLM inference.frozen20d ago12K411811060.66%
3D creation just got a serious upgrade. 🔥 From a simple prompt to fully interactive worlds, characters, camera work, lighting, and finally polished video. GPT-6 + Buzzy could make high-end 3D direction feel accessible to everyone.frozen29d ago11K3665060.46%
Marketing video creation is moving from weeks to minutes. The ability to generate, edit, and fix specific parts with a prompt makes the whole process much faster. Mark-to-fix is a particularly useful feature for quick iterations.frozen14d ago11K4976070.61%
just tried using ElevenAgents Architect through Cursor and it works really well. editing a voice agent from the same window as the backend code, with full context on the agent already there1d ago11K41814050.60%
AI infrastructure is becoming a routing problem. If the same model can be served by multiple providers at different prices, automatically finding a more efficient route can have a real impact at scale.frozen29d ago11K3369070.49%
The flexibility is what caught my attention here. Being able to change the tokenizer, data mix, or architecture without rebuilding the whole pipeline could make large-scale training much easier to iterate on.frozen9d ago11K4798070.64%
Typing makes us adapt our thoughts to the computer. Typeless flips that around by letting us communicate at the speed and flow of our thoughts. Bringing this experience to Linux is another step toward making voice a natural interface for computing.frozen14d ago11K45106080.62%
Ran a full editorial set through Claude today. Long exposure look, hard directional light, real grain. That was a studio day and a location fee six months ago. This was an afternoon at my desk in a hoodie. https://t.co/4gDG5Gp05gfrozen23d ago11K331180130.59%
Talk through a launch in the group, ask once, and get the landing page, deck, and social posts back. That’s the kind of demo I’d want to see on stage.frozen8d ago11K512380120.86%
The future of writing may be less about typing and more about expressing intent. Typeless makes that simple by turning your thoughts into polished writing. Just say what you mean, and let Typeless help you say it better.frozen15d ago11K116276071.44%
Good to see voice companies building the full stack instead of focusing on just one piece.frozen7d ago10K5152050.61%
The Emotion Engine is what makes CWM 1 especially interesting. Instead of just generating movements, it models an avatar’s internal state to predict how it should react, move, and interact. And doing whole body actions + hand object interactions without training for each character is a big step toward more capable AI characters.14d ago10K32410050.50%
Code is the result. The session is the source. AI agents can spend hours solving a problem, so that work shouldn’t disappear when the session ends or someone else takes over. AgentGit lets you save the session, preserve the context, and continue from where the work stopped.frozen21d ago9.9K511510090.86%
YOUR IPHONE HAS A HIDDEN DIAGNOSTIC TOOL MOST PEOPLE DON’T EVEN KNOW EXISTS. It can help uncover battery problems, hardware issues, camera faults, and other hidden errors in just a few minutes. Here are the codes you need:3d ago8.6K78321201332.96%
STOP using Claude just to rewrite your resume. I got 20+ recruiter calls in ONE WEEK after improving how I presented myself online. If I were actively job hunting today, I’d take it one step further: Create a 30-sec VIDEO RESUME without filming myself every single time. @veedstudio Here’s the Claude + VEED workflow 👇frozen7d ago8.1K472117091.17%
We’re a small team building on the ElevenLabs API, and V4 gives us studio-quality voice output without making our stack more complicated. That opens up a whole new level of what we can deliver to customers.frozen10d ago8.0K9818100111.70%
Apex's NanoChat implementation carries sparsity from lookup through gradient update. Making the entire computation respect a model's sparse structure feels like a broadly useful engineering instinct.frozen1mo ago7.4K3769070.79%
The future of AI video creation is conversational. Instead of learning complex editing workflows, you can simply describe the idea, review the output, give feedback, and keep refining it. Pexo turning that entire process into a conversation is a great example of how AI can make video creation much simpler.frozen16d ago7.3K3596060.77%
Big visions need bold believers. Flama just raised $40M in Series B. A new chapter, a bigger runway, and even bigger ambitions ahead. The journey is just getting started.22d ago7.0K3281050.66%
This is where AI video creation is heading. No complicated timelines or editing skills. Just describe what you want, give feedback, and let the AI handle the production. Pexo turning a simple conversation into a polished SaaS video is a great example of how much easier creative workflows can become.frozen16d ago6.2K410020.11%
Parallel runs across both fold states is the right call. If you're testing a foldable serially you're already behind.1d ago5.8K32296081.28%
Big visions need bold believers. Flam just raised $40M in Series B. A new chapter, a bigger runway, and even bigger ambitions ahead. The journey is just getting started.frozen21d ago5.7K1530090.47%
K2-Horizon-36B-A4B shows an interesting way to scale attention more efficiently. MoVA activates only ~4B parameters per token while keeping performance close to the 32B dense model across several benchmarks. The 50% reduction in gradient clipping during training makes this approach even more interesting.frozen16d ago5.4K4097051.13%
It’s amazing how they are building around bringing context and intelligence together to create a more personalized and proactive AI experience.frozen29d ago5.2K1831070.56%
MOST PEOPLE USE CLAUDE LIKE GOOGLE. THE SMART ONES USE IT LIKE A 1-ON-1 PERSONAL TUTOR. HERE’S THE PROMPT SYSTEM 👇21d ago1.7K221320112.84%
ChatGPT Prompts to Learn Anything 10X Faster Follow @techwithakansha for more https://t.co/hnATTzjL6ffrozen26d ago1.6K361410204.36%
TEACHERS, STOP SPENDING HOURS MAKING LESSON MATERIALS. 🤯 ChatGPT can turn almost ANY lesson into engaging classroom visuals in seconds. Flashcards, diagrams, posters, real-life scenarios, activities and more. Here are 10 powerful prompts that can completely change the way you teach 👇👇17h ago1.2K33163034.73%
🚨 CLAUDE CAN NOW HELP YOU BUILD A YOUTUBE CHANNEL FROM ZERO. No team. No expensive tools. No complicated strategy. Just 7 prompts to plan, create, grow, and work toward monetization in 90 days. And you can start for FREE. Here are the 7 prompts: 👇frozen9d ago1.0K28134075.12%
Claude can now be your 24/7 AI language tutor. 🌍 Learn to SPEAK a new language in just 4–6 weeks—for FREE. Here are 8 prompts that can help you learn almost any language 👇 Save this. Your personal AI tutor is one prompt away. 🔖frozen14d ago83717114024.06%
Faceless YouTube is quietly becoming a HUGE opportunity. No DSLR. No studio. No face. Just: 🧠 Claude for scripts 🎨 Google Flow for scenes ✂️ CapCut for editing 📹 YouTube for distribution Full step-by-step + prompts 👇22d ago6251762044.64%
GOODBYE, CANVA. 🚨 CLAUDE CAN NOW HELP YOU BUILD A COMPLETE BRAND KIT LIKE A $5,000 DESIGN STUDIO, WITHOUT THE $5,000 PRICE TAG. HERE ARE 8 PROMPTS TO BUILD IT STEP BY STEP, IN THE RIGHT ORDER:frozen10d ago5412557067.95%
AFTER USING CLAUDE FOR 2 YEARS, I CAN CONFIDENTLY SAY THIS: IT HAS COMPLETELY CHANGED THE WAY I WORK, LEARN, AND MANAGE MY DAY. HERE ARE 15 PROMPTS I USE ALMOST DAILY THAT HAVE MADE A HUGE DIFFERENCE IN MY PRODUCTIVITY AND WORKFLOW. THEY MIGHT DO THE SAME FOR YOU. SAVE THIS 🔖frozen13d ago533321031.69%
When you find out you can co-create any design according to your own preferences https://t.co/XXgD0N6EWp17h ago4011220003.49%
Get ready for a dose of pure happiness—this thread is full of warmth! 🧵 📷 (Click in for a smile!) https://t.co/a2DiUqAUikfrozen17d ago390521012.31%