Proposition:
$f: [0, \infty) \to \mathbb{R}$ be a continuous function and suppose that $f(0) < 0$, then $\exists \alpha > 0, \text{s.t. } f(t)<0, \forall t \in [0, \alpha] .$
Sketch:
Proof attempt:
Since $f$ is continuous, therefore for all $t, t^\prime \in \mathbb{R}_{\geq 0}$, $\forall \epsilon >0, \exists \delta >0$ such that $|t^\prime - t| < \delta \implies |f(t^\prime) - f(t)| < \epsilon$.
Then take $t^\prime = 0$, we have for all $t$, $|t| < \delta \implies |f(0) - f(t)| < \epsilon$.
Take $\delta = \alpha$, then $|f(0) - f(t)| < \epsilon, \forall t \in (-\alpha, \alpha)$.
Since $|f(0) - f(t)| < \epsilon \Longleftrightarrow -\epsilon < f(0) - f(t) < \epsilon$, therefore $f(t) < f(0) + \epsilon$.
Thus for any $0< \epsilon < |f(0)|$, $f(t) < 0$ for all $t \in (0, \alpha)$.
There are couple problems in the proof:
how do I argue that $f(t) <0$ for a closed interval as in the proposition?
how do I go from $|t| < \alpha \Longleftrightarrow -\alpha < t < \alpha$ to $t \in (0, \alpha)$?
Please let me know if there are any other problems in the proof, feel free to post yours if you can construct a more concise one.

Your two questions can be simply addressed by just observing that you can shrink the range that $t$ can be in however you want. For instance, if you know that $f(t)<0$ for all $t$ in $(-\alpha,\alpha)$, you also know that $f(t)<0$ for all $t\in[0,\alpha)$, since any $t\in[0,\alpha)$ is also in $(-\alpha,\alpha)$. (Actually, note that you can only use values of $t$ which are in the domain of the function in the first place, so you actually only know that $f(t)<0$ for $t\in [0,\alpha)$, since for $t<0$ it is undefined.) You can also use this idea to get a closed interval: can you think of a closed interval $[0,\beta]$ that would be contained in $[0,\alpha)$? The answer is hidden below.
There are some other issues in your proof. First, your definition of continuity is incorrect. Your definition would be true of any function at all, since you could just choose $\delta$ to be less than $|t-t'|$ so that the implication $|t^\prime - t| < \delta \implies |f(t^\prime) - f(t)| < \epsilon$ is vacuously true. The "for all $t$" in the statement needs to come after the "there exists $\delta$" so that $\delta$ can't depend on $t$.
Also, beware that your choice of $\delta$ (and therefore $\alpha$) depends on $\epsilon$. So you shouldn't really be talking about $\delta$ until you've chosen the $\epsilon$ you're going to use, as you do when you say $0<\epsilon<|f(0)|$. Or, if you want to leave the choice of $\epsilon$ until the end, you should state explicitly that you are using an $\epsilon$ that will be chosen later.